Post on 12-Sep-2019
Technische Universität München
Physik-Department
Lehrstuhl für Biophysik E22
Forces, Thermodynamics and Structure of Artificial Glycocalyx Models in Two and
Three Dimensions
Matthias F. Schneider
Vollständiger Abdruck der von der Fakultät für Physik der Technischen Universität
München zur Erlangung des akademischen Grades eines
Doktors der Naturwissenschaften (Dr. rer. Nat.)
genehmigten Dissertation.
Vorsitzender: Univ.-Prof. Dr. M. Kleber
Gutachter: 1. Univ.-Prof. Dr. E. Sackmann
2. Univ.-Prof. Dr. J. Friedrich
Diese Dissertation wurde am 22.04.2003 bei der Technischen Universität München eingereicht und durch die Fakultät für Physik am 06.06.2003 angenommen.
Danke an �
� Prof. E. Sackmann für die exzellenten Vorraussetzung in seinem Labor
� Dr. M. Tanaka, der mir die Freiheit ließ, eigene Ideen zu verwirklichen ohne
sein Begeisterung an meiner Arbeit zu verlieren
� Prof. R. R. Schmidt und Christian Gege für ihre ausgezeichneten Synthesen
� Prof. G.G. Fuller und seiner Arbeitsgruppe in Stanford für ihr wissenschaftliches
�know how� und die Gastfreundschaft
� PD Dr. A. Boublicht und Prof. D. Andelman deren theoretische Überlegungen
zu meinen experimentellen Resultaten zu einem tiefergehenden Verständnis
beitrugen
� Dr. Michael Rappolt für die Kooperationsbereitschaft und die Berechnung von
Elektronendichteprofilen
� PD. U. Rothe und Dr. G. Bendas für ihre Flusskammerexperimente
� Gerald Mathe und Florian Rehfeld für die Einführung und Unterstützung bei der
Ellipsometrie
� den ehemaligen Ulf Rädler, Julia Nissen und Heiko Hillebrandt für die vielen
kleinen Tips und Tricks beim Präperieren und Umgang mit Chemikalien, sowie
Roman Zantl und Frank Artzner für ihre theoretische wie praktische Hilfe bei der
Röntgenstreuung
� Zeno Gutenberg der stets seine gesammelte Erfahrung zur Verfügung stellte
und immer Zeit fand über grundsätzliche biophysikalischen Fragenstellungen zu
diskutieren
� Laurent Limozine der mir die Präparation von riesen Vesikeln beibrachte und
mir mit seiner Erfahrung am Mikroskop bis in den späten Abend zur Verfügung
stand
� Oli Purucker und Klaus Adelkofer für ihre Organisation und Hilfsbereitschaft
sowie den Rest der Tanaka Gruppe Stefan, Uwe (alias Murrat) und Jockey (alias
Joachim) für die gute Atmosphäre und ausserlaborlichen Aktivitäten
� unseren Werkstattleitern Erwin und Rudi die durch ihre Ideen und Kompetenz
erst zum gelingen vieler Experimente beidrugen
� alle übrigen Mitglieder des Lehrstuhls E22, die für das ausgezeichnete
Arbeitsklima im Institut maßgeblich waren
� meinen Bruder Stefan für seine Unterstützung nicht nur bei medizinisch
relevanten Fragen sowie seiner Gattin Birgit in deren �Tegernseeer Häusle� ein
Großteil dieser Arbeit verfasst wurde
� meiner Freundin Vanessa für ihre vielen sprachlichen Korrekturen und Tips,
aber v.a. für die nötige moralische Unterstützung
� meinen Eltern ohne deren Unterstützung und Vertrauen es nie zu dieser Arbeit
gekommen wäre
Meinen Eltern Afra und Theo
1 SUMMARY 1
2 INTRODUCTION 5
3 MATERIAL AND METHODS 11
3.1 Film balance and Langmuir-Blodgett-Technique 11 3.1.1 Physical principles of the film balance technique 12 3.1.2 The design of the trough with Wilhelmy plate 13 3.1.3 Fluorescence Film balance 14 3.1.4 Film Preparation and Langmuir-Blodgett transfer 15
3.2 Ellipsometry 16
3.3 Interfacial Rheology 21 3.3.1 Theory of Surface Rheology and measuring principle 21 3.3.2 Experimental Setup 22
3.4 Differential Scanning Calorimetry (DSC) 24 3.4.1 Theory of Calorimetry 24 3.4.2 Experimental Setup 25
3.5 X-Ray Scattering 26 3.5.1 Physical Principles of X-Ray Scattering 27 3.5.2 Experimental Setup 29
3.6 Chemicals and Chemical Structures 30
4 RESULTS AND DISCUSSION 33
4.1 Glycolipids with Linear Head group Conformation (Lac1-3) 34 4.1.1 Pressure Area Isotherms 34
4.1.2 Swelling Behaviour of glycolipid monolayer 38 4.1.3 Rheology at the Air/Water Interface (Schneider, Lim et al. 2002) 43 4.1.4 Calorimetry and X-Ray Scattering on Glycolipid Dispersions 49 4.1.5 Summary 60
4.2 Glycolipids with Branched or Bent Head Group Conformation (Lewis X, Gentiobiose) 61 4.2.1 Monomolecular Films of Gentiobiose Lipids. 62 4.2.2 Monomolecular Films of Lewis X Lipids. 66 4.2.3 Summary 70
4.3 Phase Behaviour of Fluorinated Lipids and Artificial Microdomains 71 4.3.1 Stripe-like Phase Formation in Fluorinated Lipid Monolayer (Schneider, Andelman et al. 2003)
72 4.3.2 Design of Artificial Glycolipid Microdomains by Fluorinated Lipids (Gege, Schneider, et al
2003) 81 4.3.3 Summary 89
5 CONCLUSIONS AND OUTLOOK 90
A. Viscoelasticity of PEG-lipids 93
B. Preparation of Giant Unilamellar Vesicles (GUVs) 95
Summary
1
1 Summary
The glycocalyx, a network of oligo- and polysaccharide chains with glycolipids,
glycoproteins, and proteoglycans on the extracellular membrane surface serves as
a hydrophilic �cushion� between cells in addition to stabilizing the structure of
animal cell membranes by a combination of various physical forces (generic
interactions). Furthermore, it contains specific recognition sites for counterpart
lectins and cell adhesion receptors (specific interaction). The interplay of these
generic and specific interactions often mediates adhesion and recognition between
Summary
2
cells, in which the condensation of cell surface receptors builds functional
microdomains, which can serve as a prerequisite for cell contact formation.
Although such phenomena have been widely studied, the physical basis of
glycocalyx function has not yet been experimentally understood and still little is
known regarding the interaction mechanisms on a molecular level. Despite a
number of reports on the effects of ethylene glycol chains (as a glycocalyx model
system) on the morphology and interfacial properties of membranes, studies of the
thermodynamic and elastic properties of glycolipids themselves in a well-defined
artificial model system are still missing. Consequently, a set of synthetic glycolipids
with various carbohydrate head groups and lipid anchors (synthesized by Dr. C.
Gege and Prof. R.R. Schmidt, Universität Konstanz) was used to study the
mechanical, morphological and thermodynamic properties of glycocalyx model
systems and the relationship to the molecular structure of these glycolipids.
Furthermore, the formation of functional microdomains, important for specific
interactions between membranes was studied using synthetic glycolipids with
biofunctional relevant head groups.
In Chapter 4.1.1 and 4.1.2 thermodynamic phase behaviour and hydration forces
in synthetic glycolipid monolayers were measured by a combination of Langmuir
film balance experiments and ellipsometry under controlled humidity conditions. As
model systems for the study of the impact of saccharide chain length, synthetic
lipids with linear oligolactose head groups were used. Thermodynamic parameters
such as phase transition entropy and latent heat could be quantitatively estimated
by the application of the Clausius-Clapeyron equation and were found to be
comparable to those of phospholipids. Under controlled humidity conditions the
strength of hydration (disjoining pressure) could be precisely set, yielding to
quantitative force-distance relationships perpendicular to the membrane surface.
The hydration of the oligolactose chains could be treated within the theoretical
framework of polymer �brushes�. Analysis within this framework, as well as
examination of the film balance experiments, indicates increasing entropic
contributions from the head group as a result of the elongation of the saccharide
chain. The monomolecular film of elongated sugar chains can therefore be viewed
as a �soft cushion�.
Summary
3
To get deeper insight into the relationship between molecular structure and
macroscopic physical properties (i.e. chain melting, lamellar spacing, lateral
packing and degree of hydration) the thermotropic phase behaviour of oligolactose
lipids were studied by a combination of differential scanning calorimetry (DSC) and
small and wide angle x-ray scattering and are presented in Chapter 4.1.4. The
hydrophobic/hydrophilic balance (competition between enthalpic contribution from
chain-chain interaction and entropic contribution from carbohydrate-carbohydrate
interaction) was found to be crucial in determining the morphology of glycolipid
membranes with oligolactose head groups. This dominate effect of the
hydrophilic/hydrophobic balance can be attributed to the small sterical mismatch
between the alkyl chains and the linear, cylindrical oligolactose head groups.
Presented in Chapter 4.1.3 are the studies of the mechanical properties of the
same glycolipid monolayers at air/water interface using a quantitative interfacial
stress rheometer (ISR) under well-defined thermodynamic conditions
(temperature, area per molecule, surface pressure). The hydrophobic/hydrophilic
balance was found to significantly influence the viscoelastic properties of glycolipid
monolayers. For the lipid with trilactose head group, a rheological transition (from
viscous to elastic) due to the exclusion of the hydrating water and the formation of
a physical network of hydrogen bonds could be observed. Lipids with shorter
oligolactose head groups did not display this rheological transition, revealing the
critical role of the number of hydrogen bonding sites on network formation
(cooperative effect). This approach was extended for the examination of another
glycolipid with the same monosaccharide composition as lactose but different
glucosidic bond. The results demonstrated the strong influence of steric (i.e.
entropic) contributions from the carbohydrate head group on the viscoelasticity of
the monolayer (Chapter 4.2.1). Furthermore, lipids with the asymmetric head
group Lewis X, displayed an isotropic-to-nematic transition of their lipid head
groups (Chapter 4.2.2), which could not be resolved in Langmuir isotherm
experiments.
In an additional model system, micro-domains of glycolipids with specific functions
(e.g. blood group antigens sialyl-Lewis-X) were designed by introduction of
Summary
4
partially fluorinated lipid anchors (F-alkyl chains). These lipids demonstrate
strong de-mixing with alkyl chain lipids, which were used as the matrix system.
Initial studies examined the thermodynamic properties of the pure F-alkyl lipid
monolayer (Chapter 4.3.1). Fluorescence microscopy revealed a stripe like phase
which can be explained theoretically as a consequence of the strong dipole
moment of the terminal �CF3 group observed by surface potential measurements.
The unique phase behavior of partially fluorinated lipid anchors, which results in a
strong de-mixing with matrix phospholipids, was utilized to confine functional
carbohydrate ligands within micro-domains (Chapter 4.3.2). Fluorescence
microscopy and laser scanning confocal microscopy showed that it is possible to
self-assemble so-called �artificial lipid rafts� both in monolayers and in giant lipid
vesicles. Additionally, dynamic flow chamber experiments demonstrated that the
size and distribution of functional rafts can strongly influence the dynamic cell
adhesion under shear flow.
Introduction
5
2 Introduction
Biological membranes maintain the essential differences between the cytosol and
the extracellular environment and between the contents of each cell organelle and
the cytosol. The general structure common to all biological membranes is a very
thin film (approximately 5 nm) of lipids and proteins, held together primarily by
noncovalent interactions. The proteins stick in the lipid membrane or protrude
through it and have a huge variety of different functions, ranging from the transport
of specific molecules, sensory functions and cell signalling to their actions as
catalysts in membrane-associated reactions, such as ATP synthesis, or work as
Introduction
6
structural links that connect the cytoskeleton through the lipid bilayer to the
extracellular matrix or adjacent cells. Figure 2-1 shows a sketch of a cross section
of the plasma membrane of an erythrocyte. The fluid crystalline nature of the lipid
bilayer and the fact that it is a multiparticle system with collective phenomena, e.g.
self assembling, is very appealing from a physical perspective. As a consequence
of this property, the membrane exhibits some very unique features, which are of
basic importance to life. An example of this versatility is the ability of biological
membranes to effectively form organelle compartments within the cytosol while at
the same time maintaining the form variability of the outer cell membrane to fit into
and through biological tissue. Due to its fluid crystalline character, the lipid
membrane can be described using the thermodynamics of 2 D liquids. Such fluid
crystals can appear in a variety of different phase states, such as isotropic,
nematic, solid etc., displaying entirely different physical properties (optical,
electrical, mechanical etc.). This structure-function relationship and the related
phase diagram may lead to new applications in areas such as modern material
science or biotechnology. In addition new developments in drug delivery systems,
artificial implants or artificial membranes as drug testing kits can be envisaged.
Moreover, the chiral character of lipids can introduce very unique physical
properties, such as the piezoelectrical-like effect found for smectic C liquid crystals
(Brand and Pleiner 1984). With respect to biomembranes, it is known that the
physical state of the membrane can be coupled to the function of enzymes or
proteins. One example of this is the activity of the enzyme phospholipase A2
GlycocalixBand 3Glycophorin
AnkyrinSpectrin Cytoskeltons
Binding Pocket
Figure 2-1 Cross section of an erythrocyte plasma membrane.
Introduction
7
which is increased when the membrane is found in the phase transition region
(Burack, Yuan et al. 1993). Another example is the protein kinase C which
becomes more active when the membrane forms a cubic phase (Giorgione, Huang
et al. 1998). If this is an equilibrium process, thermodynamics predicts that, in the
same way as the membrane state influences the enzyme activity, the enzyme
activity can change the membrane�s phase state. As a result the morphology and
thermodynamics of membranes is a field of intense study (Schneider, Marsh et al.,
1999).
The present work examines in particular the influence of the glycocalyx on these
properties. The glycocalyx is a major part of the outer cell membrane consisting of
lipids with carbohydrates as hydrophilic head groups (glycolipids), whose
distribution is extremely asymmetric (Figure 2-1 and electron micrograph in Figure
2-2). Essentially all glycolipids are found on the extracellular side of the membrane
and can interact and self assemble by means of van der Waals forces between
their hydrophobic tails as well as hydrogen bonds between their head groups. The
glycocalyx plays fundamental and essential roles in cell-cell and cell-matrix
interactions. It serves not only as a soft cushion between cells due to its unique
swelling behavior, but also contains specific recognition sites for counterpart
lectins and cell adhesion receptors (Curatolo 1987; Hakomori 1991; Geyer, Gege
et al. 1999; Schneider, Mathe et al. 2001). In addition to carbohydrate-protein
interactions, it has been demonstrated, that cell surface carbohydrates can
selectively bind to complimentary carbohydrates of other cells (Hakomori 1991).
Furthermore, these carbohydrates stabilize the outer leaflet of the plasma
membrane of animal cells via a combination of various physical forces (e.g.
electrostatic, van der Waals, hydrogen bonding etc.) (Gabius and Gabius 1997).
Although such phenomena have already been studied, the physical basis of
glycocalyx function, as well as the interaction mechanisms on a molecular level
was not understood.
Introduction
8
This was partially due to the lack of a systematically varied set of synthetic
glycolipids until recently. Since the interaction between sugars change
dramatically with size, bond angle and charge, it is difficult to draw any
unambiguous conclusion from the experiments on just one glycolipid. Therefore, in
order to mimic the cell surface glycocalyx, phospholipids with poly(ethylene glycol)
chains (PEG lipids) have been widely applied (Harris 1992). These are believed to
avoid non-specific adhesion on lipid vesicles or protein adsorption onto solid
supports. However, it is shown in the present work, that the complex behavior of
glycolipids and glycolipid films often depends on subtle changes in length and
orientation of their head groups. The fact that this complex behavior is not found in
lipopolymer membranes, clearly emphasizes the need for a more suitable
glycocalyx model to mimic its biologically relevant functions. Moreover, for the
understanding of the physics of the glycocalyx one has to examine both the
Glycocalix Cytosol Nucleus Plasma Membrane
200 nm
Figure 2-2 (Upper graph) Electron micrograph of a lymphocyte (taken from (Alberts, Bray et al. 1994)) showing the carbohydrate layer (black). (Lower graph) 3D sketch of the outer lipid monolayer including some glycolipids (green).
Introduction
9
microscopic as well as the macroscopic properties to find the relationship
between physical properties (and hence function) and structure. Structural
changes were observed by x-ray scattering experiments and changes in physical
properties (e.g. response functions) by means of Differential Scanning Calorimetry
(DSC), film balance and viscoelastic measurements. As a result of these studies,
insight into the nature and strength of the forces acting perpendicular to, as well as
in the membrane plane was gained.
Another important role of glycolipids is their ability to form and stabilize laterally
organized functional lipid microdomains (e.g. rafts), which can be thought of as a
transient phase separation into condensed domains of lipids inside the fluid
bilayer. As a result of this condensation, the domains are slightly thicker than the
fluid matrix surrounding them, allowing the communication between the outer and
inner monolayer, which usually move independent from each other, and enabling
the accommodation of certain proteins and binding sites important for cell
adhesion (Simons and Ikonen 1997; Jacobson and Dietrich 1999). The binding
interaction can exist in the form of protein � protein, protein - glycolipid or
glycolipid � glycolipid. Examples for the latter two are the selectin-sLeX and the
homolytic LeX-LeX binding, both of which were shown to be cooperative
(multivalent binding) [Rosenberg, 1997 #1342; Varki, 1997 #1388; Welply, 1994
#1401] (Geyer, Gege et al. 1999; Geyer, Gege et al. 2000) [Hernaiz, 2002 #51; de
La Fuente, 2001 #1127; Tromas, 2001 #1384]. Although studies clearly
demonstrated that clustering of glycolipids (e.g. sLeX) appears to be an essential
prerequisite for the cell contact to take place (Vogel, Bendas et al. 1998), the
number of reports on adhesion and rolling kinetics are still few. Taking this into
account raft-like domains were reconstituted in lipid vesicles [Ahmed, 1997 #1073;
Brown, 1998 #1103; Schroeder, 1994 #1358] and in solid supported membranes
(Dietrich, Bagatolli et al. 2001; Dietrich, Volovyk et al. 2001) . However, this
allowed only little control of size and distribution of the functional domains. The
strategy developed and applied here not only overcomes these problems, but also
allows for the first time, the study of the effect of dynamic accumulation of lipids
into clusters on membrane contact (adhesion) formation. A fluorinated lipid tail
plays the key role in this approach.
Introduction
10
Fluorinated lipids are believed to be potential candidates for a variety of
applications, e.g. drug delivery systems, microdomains etc.. This is due to the high
electronegativity of fluorine (especially with respect to hydrogen), which causes
strong dipole-dipole repulsion when oriented parallel, and because of its hydro-
and lipophobicity (Riess and Greiner 2000; Riess 2002), which builds the basis for
the design of small lipid domains when mixed with hydrocarbon chains. In the last
chapter of this thesis (4.3) several studies on different lipids with partially
fluorinated anchors (F-alkyl chains) used as a single component system as well as
mixed with alkylated matrix lipids are presented. Phase separation in lipid
monolayers and vesicles was investigated by film balance experiments and
different types of fluorescence microscopy (epifluorescence, inverted and Laser
Scanning Confocal Microscopy). The microdomains designed at the air/water
interface were transferred onto a hydrophobized solid support and were subjected
to a flow of cells in flow chamber experiments, where dynamic adhesion could be
confirmed.
Material and Methods
11
3 Material and Methods
To measure forces, thermodynamics and structure of glycolipids in two and three
dimensions, the following set of methods was used.
3.1 Film balance and Langmuir-Blodgett-Technique
In order to study the macroscopic as well as the microscopic two dimensional
phase behavior of monomolecular thin films, the film balance technique according
Material and Methods
12
to Wilhelmy in combination with a Langmuir trough (Gaines 1966) was used.
3.1.1 Physical principles of the film balance technique
Using a film balance the lateral pressure π as a function of area per molecule A is
measured in order to characterize the surfactant monolayer. π is measured as the
difference in surface tension in the absence (γ0) and presence (γ1) of surfactant at
the surface (Gaines 1966)
10 γγπ −= . Eq. 3.1
It terms of energy, the surface tension is the surface free energy necessary to
create the air/surfactant monolayer. Measuring π versus A, the so-called
pressure-area isotherms, one can determine the isothermal compressibility κT by
building the differential expression
TT
AA
∂∂−=π
κ 1: . Eq. 3.2
Using the same setup, the isobaric thermal expansivity can by determined from
the area/temperature experiments
ππα
∂∂=TA
A1: . Eq. 3.3
The isothermal compressibility can be related to the surface free energy by
1
2
21−
∂∂=
TT A
FA
κ . Eq. 3.4
I.e. in all cases one measures derivatives of the free energy. Changes in the slope
of the π-A isotherms indicate phase transitions of first (horizontal) and second
order (kink). Whereas jumps in the compressibility give information about the
symmetry of the second order phase transition (Albrecht, Gruler et al. 1978), the
Clausius-Clapeyron- equation allows for the calculation of the heat of transition in
first order transitions
Material and Methods
13
( )LCLET
T AAdT
dTQ −⋅
⋅=∆
π . Eq. 3.5
Where ALE and ALC are the area per molecule in the liquid-expanded and liquid-
condensed phase respectively, and πT the lateral pressure in the coexistence
region.
3.1.2 The design of the trough with Wilhelmy plate
In Figure 3-1 the overall design of a film balance is shown. After the film is spread
on the surface and is equilibrated, the barrier is moved to compress the film. The
Wilhelmy- Plate is then dragged by the surface tension of water towards the
air/water interface. The elongation of the spring in the Wilhelmy-System is
measured by inductivity. Knowing the spring constant and the geometry of the
plate the surface tension can be calculated. Besides surface tension there is also
buoyancy and gravity acting on the plate.
ghabbamgF ραππ −++= cos)(2)( Eq. 3.6
α is the contact angle formed by the water film with the plate, b the width and a the
thickness of the Wilhelmy-Plate, h the height of the plate covered with water and
ρ the density of water. When amphiphilic molecules acting at the air/water
interface, the surface tension is reduced and hence the lateral pressure increased.
Since all measurements are taken with respect to the free water surface, the exact
Film Balance (covered)
Monomolecular Film
Barier to Compress Film
Wilhelmy System
Figure 3-1 Design of the Langmuir-Trough used. The surface tension was measured using a Wilhelmy plate. Compressing speeds vary between 20 � 100 µm/s.
Material and Methods
14
amount of subphase volume is not of crucial importance for π-A isotherms. To
control the temperature, a coil heater connected to a heating bath (Julabo,
Seelbach, Germany) was imbedded in the Teflon block, at the bottom of the
trough. Furthermore, the whole setup was kept under a flow box to avoid dust
adsorbing at the air/water interface.
3.1.3 Fluorescence Film balance
To get a close look at the phase behavior of two dimensional thin films, a
fluorescence film balance was used similar to the one developed by Lösche et al
[Lösche, 1983 #383]. A fluorescent dye used as a probe was incorporated into the
film and the lateral dye distribution was measured from the fluorescent pictures. In
these studies, the contrast in fluorescence signal was obtained by using dyes with
different solubility for liquid expanded and liquid condensed phases. These two
phases are separated by a first order phase transition. Furthermore, there is a first
order gas to liquid transition which can not be resolved in the π-A-isotherms
described in the last paragraph. The dyes used are presented in Figure 3-17. In
Figure 3-2, a schematic picture of the epifluorescence setup is shown. The
monolayer was illuminated from the top with monochromatic light (λ = 546 nm)
from a mercury lamp using a 40x LDW plan (Olympus, Hamburg, Germany)
objective. The light emitted by the fluorescence probe was registered by a SIT
camera (Hamamatsu, Herrsching, Germany) and recorded by a VCR. To digitalize
the recorded film, we used the open source software NIH image (NIH, USA).
Material and Methods
15
3.1.4 Film Preparation and Langmuir-Blodgett transfer
The mixture of lipid, and when necessary the fluorescence probe (ratio 1000/1),
was dissolved in a chloroform/methanol/water (65/25/4 vol%) solution, and spread
directly onto the air-water interface. After evaporation of the solvent (30 min.) the
film was compressed at a rate of approximately 1 Å2/sec and molecule.
Using the Langmuir-Blodgett-Technique, monolayers of amphiphilic molecules can
be transferred onto solid substrates (solid supported membranes) (Gaines 1966).
In Figure 3-3 the transfer process is sketched for the case of a hydrophilic
substrate. After the monolayer is compressed to a certain pressure, the substrate
is slowly lifted while the pressure is kept constant and the transferred area is
monitored. The result is a monomolecular film of lipids with their hydrophilic head
groups facing the substrate. With the same technique a hydrophobic substrate can
be used, resulting in a monolayer with the head groups facing the air. Prior to
Light Source(Mercury Lamp)
Detector(SIT -Camera)
Water
Air
Objective
Dichroic Mirror
Barrier
Monomolecular Film
Wilhelmy Plate
Probe
Figure 3-2
Schematic overview of the fluorescence film balance used. Green light from a mercury lamp is used to excite the fluorescence probes (filled circles) at the air water interface and the emitted light is detected by a SIT camera. The good solubility of the probes in liquid domains allows for the optical contrast between different domains.
Material and Methods
16
transfer, the substrate was cleaned in a solution of hydrogen peroxide and
sulfuric acid (piranha) for one hour and rinsed intensively afterwards.
3.2 Ellipsometry
To study the forces acting perpendicular to the membrane surface the swelling
behaviour of transferred glycolipid monolayer is investigated at controlled humidity
conditions, using ellipsometry.
Ellipsometry is an optical non-invasive technique suited for the study of the
structure and the swelling behavior of soft interfaces. With this technique the
change in elliptical polarization of the light reflected from the sample can be
measured.
Theory of Ellipsometry
Measuring near the Brewster angle, which is around 70° for our Si/SiO2/lipid
system, leads to the best thickness resolution. With the two ellipsometric
parameters, ∆ and Ψ, it is possible to determine the layer thickness or refractive
index of the deposited film. The obtained measurement is an average over the
illuminated area of the substrate (1-2 mm²). Thus, it is feasible to measure
statistical systems such as lipid films. Figure 3-4 shows the measurement principle
lipidmonolayer
Hydrophilic Substrate Figure 3-3
The Langmuir-Blodgett technique. After compression the substrate is lifted up at constant pressure. The hydrophilic head groups get physisorbed on the surface.
Material and Methods
17
of the PCSA ellipsometer (Polarizer-Compensator-Sample-Analyzer) used in
this study. A monochromatic, collimated light beam passes a polarizer of well-
defined orientation, hits the multi layer structure at an angle Φ, and is reflected at
the same angle. Then, the reflected light is detected by a rotating analyzer.
According to the Fresnel reflection equations, the incoming light is reflected at
each interface as illustrated in Figure 3-5, depending on the state of polarization,
angle of incidence and refractive indices of the layers. With the total Fresnel
reflection coefficients Rp and Rs, the fundamental equation of ellipsometry (Eq. 3.7)
which expresses the two ellipsometric angles ∆ and Ψ in relation to Rp and Rs
[Azzam, 1977 #384] , and can be derived:
)exp(tan ∆−= iRR
S
P ψ Eq. 3.7
The ∆ values can vary from 0° to 360° and Ψ from 0° to 90°. From the equation (Eq.
3.7) one can calculate Ψ directly.
*
*
tanss
pp
RRRR
=Ψ Eq. 3.8
Figure 3-4
Principle of the PCSA ellipsometer
Material and Methods
18
For the determination of ∆ two equations are necessary:
**
*Recos
sspp
sp
RRRR
RR=∆
Eq. 3.9
and
**
*Imsin
sspp
sp
RRRR
RR=∆
Eq. 3.10
Among the two major classes of ellipsometers, the nulling ellipsometer and the
photometric ellipsometer, the one with a rotating analyzer (rA) used in this study
belongs to the second category. In this setup the intensity of the reflected light is
monitored according to the position of the analyzer. The polarity of ∆ can be
determined by two measurement cycles, one with the λ/4 - plate as compensator
and one without.
The detailed calculation of ∆ and Ψ from the measured intensity is computed by a
Fourier transformation as described in the literature [Azzam, 1977 #384]. For the
bulk silicon, a complex refractive index of n = 3.868 - i0.024 for the wavelength of
λ = 632.8 nm was assumed. The thickness and the refractive index of successive
layers were fitted from the measured ∆ and Ψ values using the �Fitpaket� program
Figure 3-5
Reflection of the laser beam at the interfaces of the multi layer structure. Ep and Es are the wave vectors parallel and perpendicular to the plane of incidence, which contains the beam and the surface normal.
Material and Methods
19
[Neumaier, 1999 #385].
The Experimental Setup
The experimental setup of the PCSA ellipsometer (Plasmos GmbH
Prozeßortechnik, München, Germany) combined with the climate chamber is
shown schematically in Figure 3-6. To adjust the angular position, a rotation stage
can be moved in x, y direction. With an autocollimator the laser light path is
adjusted to sample and detector. The film lift enables the control of the vertical
position of the substrate. Three points were measured for each sample (at 8.8 cm,
9.3 cm, 9.8 cm in the readout of the film lift control) to measure mean values for
the background data and the initial film thickness.
The conventional method for controling the relative humidity of the atmosphere is
to put a salt solution with well defined concentration in a closed chamber.
However, this requires up to 24 h to reach the equilibrium state and constant
environmental conditions (temperature, humidity and pressure) are essential.
Therefore, in this study a constant flow of air was applied. The pressurized air was
air in
air out
polarizer
analyzer
detector
He-Ne-laser
closed humidity chamber
film lift
sample
humiditymeasurement
Figure 3-6 Ellipsometer with humidity chamber and film lift.
Material and Methods
20
filtered and dried through a prefilter and an active carbon filter (Filter system
G3XA, Zander, Essen, Germany), saturated with humidity in two water baths, then
cooled through a Liebig-cooler to obtain the desired humidity inside the chamber.
The relative humidity in the measurement cell was monitored by a digital
hygrometer (Type MP100A, Rotronic, Ettlingen, Germany) in the vicinity of the
sample. This experimental setup enables the control of the relative humidity from
about 5% up to 98% [Elender, 1996 #386].
Static Swelling
Throughout the static swelling experiments, the equilibrium thickness of the lipid
was measured about 10 min after the adjustment of the humidity condition in order
to ensure thermodynamic equilibrium between the lipid film and the surrounding
atmosphere. The refractive index of the swelling polymer was adjusted by applying
the Garnet formula [Garnet., 1904 #387] (Eq. 3.11).
Φ−−
+Φ+=
)2
(
31
220
220
M
MMF
nnnn
nn Eq. 3.11
nM and n0 are the refractive indices for the pure solvent (here: water, n = 1.33) and
the solute (lipid) and Φ the volume fraction of the solute (i.e. reciprocal value of the
swelling ratio). By starting from n0, the refractive index corresponding to the initial
(dry) thickness d0 of the polymer, the apparent thickness was calculated. The
modified refractive index of the lipid head groups nF can be estimated by applying
Garnet's formula successively until the thickness and refractive index nF are self-
consistent. Absolute values of the disjoining pressure can be given as a function of
relative humidity for equal chemical potentials according to van't Hoffs law
[Landau, 1987 #388] (Eq. 3.12).
−≈−= 2
8 ln10*4.1lnmNXX
VRTp Eq. 3.12
T represents the temperature and R is the gas constant. Vm is the molar volume of
the solvent (water), and X denotes the relative humidity. The change in thickness
due to the water uptake can be normalized to the swelling ratio d/d0. The obtained
Material and Methods
21
equilibrium thickness as a function of relative humidity was analyzed in terms of
the disjoining pressure - swelling ratio relationship, i.e. a force-distance curve with
the effective force acting perpendicular to the lipid monolayer.
3.3 Interfacial Rheology
In order to study forces in the plane of a two dimensional monolayer an interfacial
shear rheometer (ISR) developed in the laboratory of Prof. G.G. Fuller (Stanford
University) was used. Rheometric measurements give insight in the side-by-side
interactions between molecules restricted to two dimensions. This is in contrast to
the out of plane forces measured in a swelling experiment in the last subsection.
3.3.1 Theory of Surface Rheology and measuring principle
In order to learn about the surface rheology of such delicate thin films like
glycolipid monolayer it is important to decouple the drag of the probe on the
surface from the subphase. The surface sensitivity is expressed as the
Boussinesq number
Where µ, µs are the subphase and surface viscosity, v is a characteristic velocity,
l, ls are the length scales in which the velocity decays in the subphase and surface
and P and A are the contact diameter and area respectively (Brooks, Fuller et al.
1999). Gain sensitivity towards the surface rheology requires B >> 1. The simplest
parameter to vary experimentally is the geometry of the probe, hence the ratio
P/A. It is minimized for the rotating disc and optimal for edge probes, since for
them P/A ~ (edge thickness)-1. For the magnetic rod used in our studies (30mm
long, 0.450 mm in diameter) B becomes 2.8 mm-1 and we conclude that the main
contribution of our detected signal results from the viscoelastic properties of the
Alv
Plv
agSubphaseDrgSurfaceDraB s
s
µ
µ== Eq. 3.13
Material and Methods
22
surface.
Applying a sinusoidal stress σ to a viscoelastic system results in a sinusoidal strain
response α with different amplitude α0 and a certain phase shift δ (Figure 3-7).
This signal response can be split up into two components. The one in phase (real
part) represents the stored energy (elastic properties) of the system while the one
with a 2/π phase shift (imaginary part) takes the part of the lost energy (viscous
properties). To summarize the two contributions into one elastic constant the
complex dynamic surface module *G is introduced
).()()( ''')(
0
0* ωωασω ωδ
ssi
s iGGeG +== Eq. 3.14
Here, 0σ is the stress amplitude, 'G the so called storage modulus and ''G the
loss modulus. From the phase shiftδ , the relationship between 'G and ''G can be
derived
'
''
tanGG=δ Eq. 3.15
By necessity ''G becomes 0 for an entirely elastic or hookean system (no phase
shift) while 0' =G for a purely vicious or Newtonian system ( 2/π phase shift).
3.3.2 Experimental Setup
A sketch of the experimental setup used to measure two dimensional
Stress
Strain
ω [rad/s]
Stre
ss, S
train
δ
Figure 3-7 Stress, Strain relationship with phase shift r during one oscillation.
Material and Methods
23
viscoelasticity is shown in Figure 3-8 and Figure 3-9. The self-built ISR is
coupled to a Langmuir film balance (KSV Instruments, Helsinki). A magnetized rod
(length L = 30 mm, diameter φ = 450 µm) resides at the air/water interface, and is
confined in a narrow channel (channel width W = 2.0 cm). By the nature of the
geometry of the channel (Figure 3-9) applied force and displacement are related
with the stress and strain amplitude as follows
LF2
�0 =σ ,
2/�
0 wx=α
A sinusoidal magnetic field gradient created by a pair of Helmholz coils was
applied to elongate the rod at a certain frequencyω , and the displacement of the
rod was monitored by a photodiode array. The translation of the rod causes a
simple shear flow to occur at the interface. The measurements were carried out at
20ºC, and the frequency of the oscillation was set constant at 1 rad/s, if not stated
otherwise. Lateral pressure or area per molecule respectively, were varied to study
the effect of surface concentration on the viscoelasticity of the glycolipid
Figure 3-8 Schematic overview of the interfacial stress rheometer (ISR).
Material and Methods
24
monolayer.
3.4 Differential Scanning Calorimetry (DSC)
To build the bridge between the thermodynamic behavior of glycolipids in two (film
balance) and three dimensions calorimetry experiments were performed. The film
balance described above was used to study phase transitions in two dimensions
and the calorimeter was applied to search for temperature induced phase
transitions in lipid vesicles, lamellar stacks of membranes etc.
3.4.1 Theory of Calorimetry
Using calorimetry the heat capacity cp of a substance or dispersions can be
determined. Since phase transitions show remarking behavior in their response
functions, calorimetry is suitable for the study of phase transitions in lipid
membranes. The heat capacity is defined as
pp T
Qc∂∂= Eq. 3.16
Where the derivative of the heat of transition Q has to be taken at constant
pressure. At constant pressure however
HQ ∆=∆ Eq. 3.17
and therefore,
Figure 3-9 Close up of the ISR from the top.
Material and Methods
25
pp T
Hc∂∂= Eq. 3.18
At the melting transition 0=∆G and since STHG ∆−∆=∆ ,
SHTm ∆
∆= Eq. 3.19
i.e. from integrating the heat capacity both the heat as well as the change in
entropy accompanied with the phase transition can be calculated, by determine mT .
Lipid membranes are known to be polymorphic systems dependent on
temperature and concentration [Seddon, 1995 #389]. Although all basic
thermodynamic quantities can be concluded from the DSC experiment the exact
structure of the corresponding phase remains unclear; this is why x-ray scattering
experiments are necessary. The relating theory and experimental setup used is
explained in subsection 3.5.
3.4.2 Experimental Setup
In Figure 3-10 a schematic drawing of the DSC used (VP-DSC, Microcal, USA) is
shown. An electronical circuit equals the temperature of both Tantal cells (sample
(1) and reference (2)) by controlling the heating (or cooling) rate. A crystal sensor
(3) measures the temperature difference between the two cells. Accordingly, the
PC (5) switchess power to the heating coils (4) to account for temperature
differences.
Material and Methods
26
The power difference multiplied by the time interval equals the heat ∆Q of Eq.
3.20. Therefore, the heat capacity for a certain time interval can be calculated by
1
1121 ))((
)(−
−−
−−∆+∆
=∆∆=
ii
iiiiiP TT
ttPPTQTc Eq. 3.20
3.5 X-Ray Scattering
X-ray scattering is one of the most powerful techniques used to determine the
periodic structure of liquid crystals in the range of a few angstroms. The small
angle region identifies the symmetry and long range order of the phase, whereas
the wide angles give information on the molecular packing, or short range order of
the phase. The changes in lattice spacing and symmetry which take place during
phase transition, can therefore be resolved and combined to the thermodynamical
properties acquired by DSC.
Figure 3-10 Schematic drawing of the DSC. Explanation in text.
Material and Methods
27
3.5.1 Physical Principles of X-Ray Scattering
According to Hyugens the outgoing wave scattered at some obstacle can be
described by a wave of spherical shape. Assuming a plane wave coming in
(Figure 3-11)
)(0
0)( trqieArE ω−−=rrrrr
a fraction f gets scattered and gives the amplitude at the point 0Rr
at a distance Rr
from the scattering center
tirqqiRqiRqi eeeRAfe
RrEfRA ω−−==
rrrrrrrrr
r)(0
000
)()(
Considering a lattice with N scattering centers the total amplitude becomes
∑ =∆= N
nrqi
nnefCqA
1)(
rr
.
Where )(0 0 tRqieRAC ω−=
rr
and qqq rrr −=∆ 0 is the difference between the incoming and
X
Y
Zr
R0
Rq0
Figure 3-11 Huygens principle. The planar and coherent incident wave 0qr is scattered at rr . This results in an outgoing spherical wave.
Material and Methods
28
outgoing wave vector. If the scattering center consist of i units (atoms,
molecules etc.) and every atom scatters a fraction if , then
))(()( SFFCeefCqAn
rqi
i
rqii
ni == ∑∑ ∆∆ rrrr
.
The form factor (F) resulting from the sum over i depends on the conformation and
scattering properties of the atoms (or molecules) inside the unit cell. However, the
structure factor (SF) resulting from the sum over all scattering centers n ,
represents the symmetry (Bravais-Lattice) of the liquid crystal. Since the measured
intensity in a scattering experiment is the square of the total amplitude the
information about the phase gets lost.
222 )()()( SFqFqAqI =∝
In mathematical terms SF is the Fourier transformation of the real lattice with its
Bragg-Peaks being the lattice point of the reciprocal lattice ),,( lkh . From this peaks
the distance between neighboring planes can be calculated using
2*2*2* )()()(22
lckbhaqd
hklhkl
++== ππ .
Where *** ,, cba are the vectors spanning the reciprocal lattice. This represents the
fact that the reciprocal lattice vectors are perpendicular with respect to their planes
in real lattice. From the form factor F the electron density profile can be calculated
by inverse Fourier transformation. Going from the sum to the integral
representation
dVezFxqi rr
−
∫= )(ρ
With )(zρ being the electron density at the point x in the volume dV . The Fourier
transformation gives
)2cos()cos()()(max
1 dhzFdVqzqFz
h
hhV
πρ ∑∫=
±≈= Eq. 3.21
h represents the order of the reflection, hF the respective form factor and d the
Material and Methods
29
lamellar spacing. Here centrosymmetry of the crystal was assumed, as found in
stacks of lamellar bilayers. As a consequence, the unknown phases are either 0°
(+) or 180° (-). From all the possible phase shift combinations the most likely is
picked to reconstruct the electron density profile.
3.5.2 Experimental Setup
Figure 3-12 shows a schematic drawing of the setup used. Wheresa the resolution
of the SAXS data is around 4 � 10 nm which therefore resolves the lattice spacing,
the WAXS data are suited for the investigation of in plane correlation in the range
of a few nm down to 1 Å. The suspensions with the concentration of 20 wt% water
were filled into quartz capillaries (Hilgenberg, Malsfeld, Germany). The
experiments were performed at three different beamlines. SAXS data were
measured at the synchrotron beamline ID2A of European Synchrotron Radiation
Facility (ESRF, Grenoble), with a resolution better than ∆q = 0.0015 Å�1. WAXS
data were taken at the beamline D43 of Laboratoire pour l�Utilisation du
Rayonnement Electromagnétique (LURE, Paris). In this case the resolution was
∆q = 0.0055 Å�1. Furthermore SAXS and WAXS experiments were done at the
beamline A2 (HASY-Lab) at DESY (Deutsches Elektronen Synchrotron) in
Hamburg. In all cases, SAXS and WAXS, the observation of isotropic Debye-
Scherrer rings indicated that all the samples consisted of perfect powders. The
radial integration of the two dimensional data recorded using the local CCD
camera at ID2A, was carried out by the standard routines of ESRF. At LURE, data
was collected using Fuji image plates in combination with homemade data
processing software on the basis of Igor PRO (Wave Metrics Inc., USA).
Material and Methods
30
3.6 Chemicals and Chemical Structures
Unless otherwise specified, all lipids (including labelled) were dissolved in a
chloroform/ethanol/water (65/25/4) solution (called �magic�). All lipids used were
synthesized by C. Gege in the laboratories of Prof. R. R. Schmidt at the Universität
Konstanz, Germany. Glycolipids with both linear and branched head groups were
used. The samples with linear head group were named Lac N, corresponding to
the number of lactose units, N = 0, 1, 2, and 3 (Figure 3-13). Details of the
synthesis have been reported elsewhere (Schneider, Mathe et al. 2001).
OC 16H33
OC 16H33O
O
HO
OH
OH
OH
OO
HOOH
OH
N = 0,1,2,3N
N = 0 = -OH-head group
Figure 3-13 Chemical structure of the synthetic glycolipids with oligolactose head groups, Lac N (N=1-3).
SAXS - Detector
WAXS - Detector
SampleBeamline
θ
r
Figure 3-12 The scattered wave of a monochromatic, coherent incident wave, appears under an angle Θ at the detector.
Material and Methods
31
The branched (bent) glycolipids studied all have the same hydrophobic tails and
glycerol junction however, the head group were either the disaccharide
gentiobiose and the trisaccharide Lewis X (plus lactose spacer) as shown in
Figure 3-14 and Figure 3-15.
O
O(CH 2)8 (CF 2)7CF 3
O(CH 2)8 (CF 2)7CF 3
OH
SLeX
LeX
Figure 3-16 Chemical structure of the per fluorinated lipid with the three different head groups used in this study.
Figure 3-14 Chemical structure of the synthetic glycolipid with the disaccharide head group gentiobiose.
Figure 3-15 Chemical structure of the synthetic glycolipid with the Lewis X head group and lactose spacer.
Material and Methods
32
Figure 3-17 Chemical structure of the fluorescence probes used. (Left) T-Red. (Right) Bodipy.
Results and Discussion
33
4 Results and Discussion
In the following subsections the results for a variety of glycolipids are presented
and discussed. The complete set of complementary experiments gives a good
phenomenological explanation for the enormous differences in viscoelasticity,
thermodynamics and structure found for these glycolipid membranes in two and
three dimensions.
Results and Discussion
34
In the last subsection novel compounds of fluorinated lipids are presented. The
appearance of modulated phases is discussed, in terms of dipolar forces
according to the theoretical work by D. Andelman (Andelman, Brochard et al.
1987), and their unique properties for the design of artificial lipid microdomains are
discussed.
4.1 Glycolipids with Linear Head group Conformation (Lac1-3)
4.1.1 Pressure Area Isotherms
For each glycolipid (the number of lactose units, N = 1, 2, 3), the pressure-area
isotherms were measured at several different temperature conditions between 283
K and 308 K. In order to eliminate hysteresis effects, the isotherms were
monitored during expansion as well as during compression.
The Langmuir isotherms of the Lac 1 lipid are shown in Figure 4-1. At T ≤ 298 K,
the isotherms exhibited no liquid expanded phase, and were dominated by the
condensation of the dihexadecyl chains from a gas phase to a liquid condensed
Area [Å2]
ALC
ALE
A'LC
Figure 4-1 Langmuir isotherms of the Lac 1 monolayer at different temperatures. The liquid-expanded (LE), liquid-condensed (LC) coexistence line was fitted by a polynomial of 4th order. The linear extrapolated lines were taken to define onset and endpoint of the phase transition ALE, ALC. To estimate the deviation in area several points were taken (ALC, ALC��).
Results and Discussion
35
phase. Such behaviour can be explained by the stiffness of the short, fully
hydrated, and stretched �rod-like� lactose moieties. At T = 303 K, an onset of a
plateau-like regime was observed, corresponding to a first order phase transition
from the liquid expanded to the liquid condensed state. Further rise in temperature
led to an increase in the transition pressure and a decrease in the coexistence
region. Such a systematic tendency coincides with the approach to a critical, or as
will be explained later rather tricritical point, which is well known from the previous
studies on ordinary phospholipids monolayers (Albrecht, Gruler et al. 1978;
Möhwald 1995).
As presented in Figure 4-2 a and b, qualitatively similar isotherms were observed
for the monolayers of Lac 2 and Lac 3. In accordance with the increase in lactose
units, a systematic increase in the transition pressure, pK, and a clear decrease in
the phase transition temperature was observed (Figure 4-2 a and b). The obtained
results suggest that the steric interactions between neighbouring lipid molecules
were dominated by the strong repulsion between the head groups. However, the
qualitative shape of the coexistence region was still dominated by the lateral
packing density of the alkyl chains, and not by the �polymer-like� effects of the
head groups. A similar tendency was observed in the previous study for the
monolayers of PEG-lipids with shorter chains (Mathe, Gege et al. 2000).
It should be noted, that the slope of the isotherms in the coexistence region
(a) (b)
Figure 4-2 Langmuir isotherms of a, Lac 2 and b, Lac 3 monolayers at different temperatures. Onset and endpoint of the phase transition area was defined as in Figure 4-1.
Results and Discussion
36
increases respectively with the increase in the size of the lactose head groups.
Such a slope in the Langmuir isotherms can be generally explained by the
stabilization of domains due to; i) small amounts of impurities (≥ 0.2 mol %) (Pallas
and Pethica 1985; Miller and Mohwald 1987), ii) �intermediate� states of the alkyl
chains (Mouritsen 1983), or iii) the strong interaction between the head groups
(Scott 1975). The first two approaches are based on non-equilibrium effects, which
do not follow the Gibbs phase rule, however, the third interpretation explains this
slope by the continuous compression of the head groups. By applying the
Clausius-Clapeyron equation
( ) )( LCLELCLE
K
AATq
AAs
dTdp
−∆=
−∆= Eq. 4.1
thermodynamic quantities such as the molar latent heat, q, or the molar transition
entropy, s = q/T, can be derived from the variation of the transition pressure with
absolute temperature, dpk/dT. Figure 4-3b shows the temperature dependence of
the latent heat, q, whose error bars are mainly due difficulties in defining the onset
and the end point of the transition, ALE - ALC. Latent heat is inversely correlated to
lactose moiety length this can be explained by the lower degree of cooperativity
due to the larger head groups.
The increase in pK as well as the decrease in ALE - ALC can be explained by the
b
Temperature [K]
a
Figure 4-3 (a) Transition pressure pk and (b) molar latent heat q of the phase transition plotted as a function of temperature T for Lac 1 (open circle), Lac 2 (closed square), and Lac 3 (open square).
Results and Discussion
37
approach towards a tricritical point (Albrecht, Gruler et al. 1978) where the first
order phase transition between the liquid expanded and liquid condensed phase
transforms into a second order phase transition of phases with different
orientation. This can be understood within the framework of the Landau-De
Gennes theory with a symmetrical Landau free energy of the form (Landau and
Lifschitz 1987)
6420 6
141
21 ηηη ECA ++=Φ−Φ . Eq. 4.2
Where ECA ,, are function of temperature and pressure and basically free to
choose, while the order parameter η is determined from the equilibrium conditions
of the system (minimization of Φ ). Since this is a system of two order parameters
(orientation and density) coupled by the fact that the volume of a lipid stays
constant during inclination, multicritical points are possible. It turns out that the
coefficient C can change its sign as a function of temperature and area per
molecule. At 0=C the critical behaviour abruptly changes from first to second
order, hence a tricritical point. As for the fluid-gas transition in the van der Waals
gas model, in the close vicinity of the tricritical point, the coexistence line can be
represented by a parabola whose vertex coincides with the (tri-) critical point. In
this regime, ALE - ALC in the Clausius-Clapeyron equation disappears and both the
transition entropy and the latent heat become zero (Möhwald 1995). The critical
temperature TC can be calculated by
2/1
2
−=
−
C
C
C
C
TTT
AAA . Eq. 4.3
Where CA is the middle point between ALE and ALC (Goldenfeld 1992). From the
approximately linear relation of ∆q vs. Tk (Figure 4-3b) another estimation of the
(tri-) critical temperature, CT can be performed ( 0=∆q ) and the two CT �s can be
compared to each other. The results are in good agreement and finally give a (tri-)
critical pressure of pC = 9 - 16 mN/m and a (tri-) critical temperature of Tc = 313 -
316 K, respectively, which are comparable to those of phospholipids with
dihexadecyl chains. Since the lateral pressure expected in lipid vesicles is above
Results and Discussion
38
25 mN/m, these bilayers would not be expected to be in any critical state. It should
be mentioned that the coupling of order parameters does not necessarily require a
tricritical point, because a critical endpoint can exist instead. Since it is not
possible to distinguish between these two phenomena with the technics applied,
mainly due to the fact that the film loses it�s stability at higher temperatures, the
prefix tri- is put in brackets.
4.1.2 Swelling Behaviour of glycolipid monolayer
4.1.2.1 Theoretical Concepts
The measured swelling curves were analyzed by applying two different physical
concepts; (i) the scaling theory by Alexander and de Gennes [De Gennes, 1976
#393] (Alexander 1977)[De Gennes, 1980 #395][Daoud, 1977 #396] and (ii) the
mean field approach (also called self consistent field SCF) by Milner [Milner, 1988
#391] [Milner, 1988 #391]. Although our experimental systems do not fulfil a
symmetrically planar situation with the polymers confined between two parallel
plates, all the equations discussed in the following chapter were applicable to the
experimental data by multiplying all theoretical expressions by a factor of unity.
Scaling Theory
A general expression for the free energy F of grafted polymers on the surface is
given by de Gennes, Alexander, and Daoud [De Gennes, 1976 #393; De Gennes,
1980 #395][Daoud, 1977 #396; Alexander, 1977 #394]
σξσ
δδ ln131245335
kTNaa
DkTD
NaDNa
DakTNF +
+
+
−
≈ −
−
. Eq. 4.4
N is the number of monomer segments with length a, and D is the thickness of the
polymer layer. δ represents the surface adsorption energy per monomer in units of
kT, while σ and ξ are the mean area per polymer and the blob diameter,
respectively [De Gennes, 1980 #395]. The first term describes the energy
necessary to confine a polymer molecule, behaving as an ideal chain inside a
blob. The second term stands for the adsorption energy of a chain on the surface,
while the third term represents the repulsion between overlapped polymer chains.
Results and Discussion
39
The fourth term describes the so-called �brush� regime and the last term
represents the translational entropy of the polymers, which can be neglected by
assuming that the alkyl chains are immobilized on the surface.
In the case of higher grafting density where the grafting distance is less than the
Flory radius, dp < RF, the polymer chain takes a �brush� conformation with the blob
diameter of ξ. The equilibrium thickness Dbrushst can be described as 32−≈ p
stbrush aNdD .
The interaction potential of the polymer Vbrushst is given by
+
≈
4745
3 74
54
stbrush
stbrushst
brushp
brush DD
DD
DdkTV
Eq. 4.5
yielding the resulting pressure of
−
≈
4349
3 stbrush
stbrush
p
st
DD
DD
dkTP
brush.
Eq. 4.6
Mean Field Theory
The mean field approach of Milner et al. [Milner, 1988 #391; Milner, 1988 #392] is
based on the terminally fixed linear chains exhibiting a high grafting density. In
contrast to the scaling approach, the quality of the solvent is not so crucial in this
treatment. When the film is compressed from the equilibrium thickness scfbrushD the
interaction energy per unit area can be described as
( )
−
+
−≈
5232
2
101
21
212 scfbrush
scfbrush
scfbrushscf
brush DD
DD
DD
wNV σπ Eq. 4.7
and the resulting interfacial pressure is given by
( )
−+
−
≈
4232
2
21
211
12 scfbrush
scfbrush
scfbrush
scfbrush
scfbrush D
DD
DD
DD
wNP σπ Eq. 4.8
Results and Discussion
40
σ = dp−2 is the grafting density, and w stands for the �excluded volume�.
4.1.2.2 Results
The glycolipid monolayers were transferred onto the substrate at T = 293 K and at
a lateral pressure of p = 25 mN/m. The grafting densities were 37 Å2 (Lac 1), 37 Å2
(Lac 2) and 40 Å2 (Lac 3), respectively. In this regime, the glycolipids are in the
liquid condensed phase, where the alkyl chains orient nearly perpendicular to the
surface. The relative humidity was varied between 30 % and 98 %, corresponding
to a change of the disjoining pressure (cf. 4.1.2.1) between 1.69 x 108 and 2.83 x
106 Pa. In Figure 4-4 a - c, the disjoining pressure is plotted versus the absolute
thickness of the swollen lactose layer. In each plot, both the experimental data
(open squares) and the two theoretical fits are presented based on the scaling
approach and the mean field model. To expose a possible power law dependence
between the disjoining pressure and thickness, all results are presented as log-log
plots. The swelling behaviour of Lac 1 could hardly be interpreted as �brushes�
neither by the scaling approach nor by mean field theory, even though the swelling
ratio of ~ 2.0 in the low-pressure regime (~ 107 Pa) is still in a plausible range as is
known from the corresponding ratios of dextran (~ 2.0) and hyaluronic acid (~ 2.7)
(Mathe, Albersdorfer et al. 1999). This observation suggests that the very short
head groups behave like �rigid-rods� but not like �polymer chains�, similar to what
has been observed for the PEG-lipid monolayers with shorter chains (Mathe, Gege
et al. 2000). This is also in good agreement with the Langmuir isotherms of Lac 1,
showing the qualitatively similar characteristics to phospholipid monolayers. At
high disjoining pressures from 2 x 108 to 7 x 107 Pa a power law ( )nddp 0∝ was
fitted to the disjoining pressure curves, yielding an exponent of n ≈ �9 (Figure 4-5).
In this high disjoining pressure regime, typical intermolecular distances, r, are
comparable to the Bohr radius (~ 0.5 Å), and the swelling is expected to be mainly
dominated by short-range repulsive interactions caused by the overlapping of
molecular orbitals. The hard core repulsion of the Lennard-Jones potential scales
as r-12. This exponent corresponds to a scaling law of ( ) 90
−∝ ddp [Israelachvili,
1992 #390], that agrees with the power law obtained from our experiments. Similar
Results and Discussion
41
power law dependencies in the high disjoining pressure regime could be also
observed for Lac 2 and Lac 3.
In Figure 4-4 b and c, the swelling behaviours of Lac 2 and Lac 3 were compared
with theoretical predictions for �polymer brushes� based on the scaling approach
and the mean field theory. Both of the �brush� models fit very well to the measured
disjoining pressure curves. We were not able to fit these curves by the
�mushroom� model (not shown), where lower surface coverage is assumed, which
is in agreement with the conditions of our preparations (i.e. high transfer pressure,
relatively short head groups). In addition, it should be noted that the thickness of
the lactose layers (≤ 3 nm) still remains far away from the basic statistical
condition, N » 1. These results also showed a good agreement with results for
PEG-lipids with longer chains (Mathe, Gege et al. 2000).
Thickness of Lactose Layer [nm]Thickness of Lactose Layer [nm]
Thickness of Lactose Layer [nm]
a. b.
c.
Figure 4-4 Absolute disjoining pressure versus thickness of the lactose layers: (a), Lac 1; (b), Lac 2; and (c), Lac 3. The measured values (open circles) were compared with the theoretical predictions based on the self-consistent-field approach (solid lines) and the scaling theory (broken lines). The swelling behavior of Lac 2 and Lac 3 could be well explained by both of the �brush� models.
Results and Discussion
42
The difference between the swelling behaviour of Lac 1 and that of Lac 2 and Lac
3 can be explained in the same way as the influence of the head groups on the
phase transition of the glycolipid monolayers. Actually, the Langmuir isotherms of
Lac 1 were dominated by the condensation of the alkyl chains at T ≤ 298 K (Figure
4-1). In accordance with the increase in the lactose units, the head groups gained
conformational entropy. The steric interaction between the neighbouring lipids is
strongly influenced by the repulsion between the head groups. Thus, the swelling
curves can be well explained by the �brush like� behaviour of the lactose groups.
~ (d/d0)-9
Swelling Ratio d/d0
Figure 4-5 Absolute disjoining pressure as a function of relative swelling ratio of Lac 1, d/d0, normalized to the thickness of �dry� layer. At high disjoining pressures from 2 x 108 to 7 x 107 Pa, a power law p ~ (d/d0)n was fitted to the disjoining pressure curves, yielding an exponent of n ≈ �9.
Results and Discussion
43
4.1.3 Rheology at the Air/Water Interface (Schneider, Lim et al. 2002)
To reveal insight on the forces acting in the monolayer plane rather then
perpendicular to it (cf. previous chapter), viscoelasticity measurements on
glycolipid monolayers were performed and are reported in the present chapter.
To ensure that all experiments were done in the linear viscoelastic regime, an
amplitude sweep was performed prior to every experiment. The amplitudes found
ranged between 20 and 100 µm. If not otherwise specified the frequency was set
to 1 rad/s (approx. 0.16 Hz). Figure 4-6 presents the storage modulus G� and the
loss modulus G�� of the Lac 1 monolayer, plotted as a function of area per
molecule. Even at a large area per molecule, the monolayer was already quite
viscoelastic. Both the storage and loss moduli of the monolayer showed a sharp
increase when compressed from the liquid expanded to the liquid condensed
phase. The increase in surface viscosity across the phase transition of the alkyl
chains was also found in previous studies
Area per Molecule [Ų]
Area per Molecule [Ų]
A
B
C
A
B
C
Figure 4-6 The storage modulus G� and the loss modulus G�� of the Lac 1 monolayer, measured at T = 20ºC. The oscillation frequency was set constant, ω = 1 rad/s. The Langmuir compression isotherm of the monolayer was given in inset, and the correspondence between the dynamic moduli and the isotherm is indicated.
Results and Discussion
44
of phospholipid monolayers (Kraegel, Kretzschmar et al. 1996; Naumann, Brooks
et al. 1999). This observation can be attributed to strong chain-chain interactions
that arise from film condensation, and dominate the in-plane correlation. The
surface storage and loss moduli of the Lac 2 monolayer are plotted versus area
per molecule in Figure 4-7. Similar to the results obtained for the Lac 1 monolayer,
G�� was larger than G� over a wide range of surface pressures. Nevertheless, both
G� and G�� values were remarkably smaller, by almost an order of magnitude, in
comparison to those of the Lac 1 monolayer, suggesting that the film is rather fluid.
This can be interpreted qualitatively in terms of the hydrophilic/hydrophobic
balance between the head group and the alkyl chains. When this balance is
shifted towards greater hydrophilicity, the cooperativity between the alkyl chains is
reduced (cf. chapter 4.1) and the head groups are more hydrated, resulting in a
rather fluid-like film. This can also be concluded from the linear behaviour of G�� in
Figure 4-9, where the loss moduli of Lac2 (square symbols) and Lac3 (triangular
symbols) are plotted as a function of frequency. This fluid like character was
observed over a wide range of surface pressures. It should be noted that the
�fluidization� observed here is not due to the relaxation of the lateral packing,
because the area per molecule of the Lac N in the liquid condensed phase
(π = 25 mN/m) is almost independent of the number of lactose units,
A = 37 ~ 40 Å2 (Schneider, Mathe et al. 2001). In fact, recent NMR studies and
molecular dynamic simulation have also shown that a linear tetrasaccharide
resembling the Lac 2 head group (lacto-N-neotetraose) takes a uniaxial, cylindrical
conformation in dilute liquid crystalline media such as phospholipid dispersions
(Rundlöf, Landersjö et al. 1998; Landersjö, Höög et al. 2000) and therefore does
not significantly affect the lateral packing of the hydrophobic chains. Furthermore,
the dynamic moduli of the Lac 0 (dihexadecylglycerol without saccharide head
groups) monolayer were too large to obtain any quantitative data, since strong
correlations between saturated alkyl chains are dominant. These results suggest a
continuous reduction in the chain-chain correlation with increases in the number of
saccharide units.
Results and Discussion
45
In comparison to the other examined monolayers, the viscoelastic properties of the
Lac 3 monolayer were strikingly different. As presented in Figure 4-8, the
monolayer became elastic at surface areas below 50 Å2, where the storage
modulus G� became larger than the loss modulus G��. This molecular area
corresponds to the surface pressure of around 6 ~ 8 mN/m in the Langmuir
isotherm (inset of Figure 4-8). The Lac 3 monolayer was viscous (G�� > G�) when
the surface pressure was less than this transition pressure (Figure 4-10). Above
this transition pressure, however, the elastic contribution became dominant
(G� > G��). Such a crossover point where G� = G�� is referred to as a rheological
transition point. If one considers the short (the stretched length of Lac 3 head
group is still less than 4 nm) and cylindrical head group of Lac 3, this transition can
obviously not be caused by a physical entanglement of the oligosaccharide head
groups, which would be in contradiction to the swelling behaviour studied in
chapter 4.1, but rather by the formation of a physical network of hydrogen bonds.
Interestingly, the surface pressure at which the rheological transition of the Lac 3
Area per Molecule [Ų]
Area per Molecule [Ų]
A
C
B
A
B
C
C
B
Figure 4-7 The dynamic moduli of the Lac 2 monolayer, measured at the same conditions as in Fig. 2 (T = 20ºC, ω = 1 rad/s). Both G� and G�� are smaller by almost an order of magnitude than those of the Lac 1 monolayer.
Results and Discussion
46
monolayer takes place (π = 6 ~ 8 mN/m) corresponds to the end point of the
coexistence of the liquid expanded and the liquid condensed phase (inset of
Figure 4-8). As described in chapter 4.1, compared to the isotherm of the Lac 2
monolayer (inset of Figure 4-7), the slope of the isotherm in the coexistence region
is apparently larger, indicating a decrease in the effective interaction between alkyl
chains. When the Lac 3 monolayer is compressed to the liquid condensed phase
(e.g. at π = 10 mN/m), the elastic contribution becomes dominant (G� > G��). In fact,
as seen in Figure 4-9 the measured G�� values are almost independent of the
oscillation frequency at π = 25 mN/m. In this regime, the hydrating water is
excluded and hydrogen bonding �bridges� the Lac 3 head groups during lateral
compression to higher pressures. The coupling between thermodynamic (alkyl
chain density) and rheological (head group cross-linking) transition is illustrated in
Figure 4-11. As the condensation of the hydrophobic chains takes place the water
is partially released from the interface, which enables the formation of a hydrogen
network with predominantly elastic properties. This is in contrast to the monolayers
of Lac 1 and Lac 2, where the phase transition to the liquid condensed phase
results in a significant increase in the film viscosity.
Area per Molecule [Ų]
Area per Molecule [Ų]
AB
C
A
C
B
C
Figure 4-8 The dynamic moduli of the Lac 3 monolayer. The measurement conditions were the same as in the previous figures (T = 20ºC, ω = 1 rad/s). The monolayer became elastic (G� > G��) at surface areas below 50 Å2.
Results and Discussion
47
Thus, it can be concluded that the rheological transition of the Lac 3 monolayer is
not caused by the correlation between the condensed alkyl chains, but by the
strong coupling between the linear hexasaccharide (Lac 3) head groups. These
results are in good agreement with our recent X-ray scattering experiments on
Lac N lipid dispersions, demonstrating that the strong correlation between Lac 3
head groups actually prohibits the endothermic transition of the alkyl chains, which
will be discussed in the following subsections in detail (Schneider, Zantl et al.
2002). Such strong carbohydrate-carbohydrate attractions between linear oligo-
and polysaccharides have also been reported for other glycolipids with
cellooligosaccharides (Hato and Minamikawa 1996; Tamada, Minamikawa et al.
1996) and cellulose (Brandrup and Immergut 1975; Hato and Minamikawa 1996;
Tamada, Minamikawa et al. 1996).
[rad/s]
~ f (ω)
~ ω
Figure 4-9 The loss modulus G�� of Lac 2 (square symbols) and that of Lac 3 (triangular symbols) at π = 25 mN/m, plotted as a function of oscillation frequency.
Results and Discussion
48
Upon compression, the viscoelasticity of glycolipids with the shorter head groups
(Lac 1 and Lac 2) exhibited a continuous increase through the condensation of
alkyl chains. Even the Lac 3 monolayer, where a clear rheological transition was
observed, revealed a continuous change in the dynamic moduli when the film was
compressed. This is in contrast to the studies by Naumann et al. (Naumann,
Brooks et al. 1999) who found a discontinuous change in the dynamic moduli
related to a �high-pressure rheological transition�, i.e. a crossing between G� and
G�� at high surface pressures π. A recent study of Ahrens et al. (Ahrens, Bækmark
et al. 2000) demonstrated that this �high pressure transition� resulted in the
formation of hydrophilic/hydrophobic nano-stripes with weakly ordered alkyl chains
and polymer head groups. Considering the longer, flexible head groups of PEG-
lipids (N = 44, 112), the rheological transition of lipopolymer monolayers seems to
follow a different mechanism (e.g. the in-plane interaction between these nano-
domains). In addition, the viscoelasticity of PEG 3, 6 and 9 lipopolymers
(Appendix) show a continuous decrease in G� and G�� with increasing length. This
clearly suggests the use of glycolipids as they were studied here rather than
Area per Molecule [Ų]
Area per Molecule [Ų]
51 Ų60 Ų
G' < G''G' > G''
Figure 4-10 The dynamic moduli of the Lac 3 monolayer. The measurement conditions were the same as in the previous figures (T = 20ºC, ω = 1 rad/s). The monolayer became elastic (G� > G��) at surface areas below 51 Å2.
Results and Discussion
49
lipopolymers for future studies on artificial glycocalyx models as used for example
in adhesion studies.
Crosslinked Network (Elastic)
Thermodynamic Transition
Coexisting Chains
Rheological Transition
Condensed Chains
Viscous/Newtonian
Figure 4-11 Illustration of the coupling between thermodynamical and rheological transition. The phase transition related condensation of the chains, helps to squeeze out the water between neighbouring head groups and therefore facilitates the formation of a physical network of hydrogen bonds with a higher elastic modulus.
4.1.4 Calorimetry and X-Ray Scattering on Glycolipid Dispersions
To study the morphology and corresponding phase behaviour of glycolipid
membranes in the lamellar phase a systematic combination of differential scanning
calorimetry (DSC) and small- and wide angle X-ray scattering experiments (SAXS
and WAXS) was performed.
All the measured DSC data (transition temperature and enthalpy) and the
diffraction peaks obtained by SAXS and WAXS experiments are summarized in
Table 1, and the details are described for each lipid in the following sub-sections.
Lac1
Results and Discussion
50
The heat capacity trace of Lac 1 is given in Figure 4-12 (upper graph), exhibiting a
sharp transition at Tt = 74 ûC and a phase transition enthalpy of ∆H = 30 kcal/mol.
A distinct and broad pre-transition peak was also observed at around T = 60 ûC.
The powder-averaged small angle X-ray scattering data at T = 20, 40, 60, and
80 ûC are summarized in Figure 4-12 (lower graph), indicating periodic 3D lamellar
structures. Across the main transition at Tt = 74 ûC, the periodicity of the low angle
spacing was changed from 68 Å (below) to 60 Å (above), suggesting the �melting�
of the dihexadecyl chains. As shown in Figure 4-12 (lower graph), the wide angle
patterns at T < Tt (T = 20 ûC) can be characterized with three pronounced
20 40 60 80
0
4000
8000
12000
Lac 1
T [ ºC]
0.60.50.40.30.20.1
20 C
40 C
60 C
80 C
SAXS WAXS
q [Å –1]
Figure 4-12 (Upper) Differential heat capacity scan of the Lac 1 dispersion (1 mg/mL) recorded at the heating rate of 20 ºC/h, exhibiting a sharp transition at Tt = 74 ûC and the phase transition enthalpy of ∆H = 30 kcal/mol. (Lower) Powder-averaged small angle X-ray scattering (SAXS) data of the lamellar dispersion of Lac 1 at T = 20, 40, 60, and 80 ºC). The lamellar spacing showed a transition between 60 ºC (dSAXS = 68 Å) and 80 ºC (dSAXS = 60 Å). Wide angle X-ray scattering (WAXS) data suggested the transition between the crystalline LC phase and the fluid Lα phase (lower right).
Results and Discussion
51
scatterings at 3.76, 4.45, and 7.50 Å. The scattering peak at 4.45 Å corresponds
to the alkyl chains in the lamellar crystalline (LC) phase with a triclinic packing
mode (Larrson 1988). On the other hand, the peaks at 7.50 and 3.76 Å can be
interpreted as the first- and the second order peaks due to the strong correlation
between dehydrated head groups (Seddon, Cevc et al. 1984; Caffrey 1987; Hinz,
Kuttenreich et al. 1991; Köberl, Hinz et al. 1998). At T > Tt (T = 80 ûC), a broad
band at about 4.57 Å could be observed, suggesting the fluid Lα phase of the alkyl
chains. Furthermore, the scattering peaks from the head group correlation
disappeared because the lactose groups were hydrated. Thus it has been
demonstrated that the Lac 1 lamellar has a transition between the crystalline LC
phase and the fluid Lα phase. However the corresponding phase transition
temperature of 74°C is apparently higher than that of other lipids with dihexadecyl
chains, such as DPPC (Tm = 41.4 ºC) and the obtained transition enthalpy (∆H =
30 kcal/mol) is larger in comparison to the sum of transition enthalpies of DPPC
from LC phase to Lα phase (i.e., LC . Lß� . Pß� . Lα), ∆H = 15 kcal/mol (Cevc 1993),
respectively. Alkyl chains of Lac 1 are strongly correlated by the very strong van
der Waals interaction, which even enable them to form crystalline-like tight packing
with almost no tilting. The additional enthalpic contribution may be due to the
hydrogen bonding between the Lac 1 head groups that are free from dipoles, in
contrast to phospholipids with P�N dipoles (Cevc 1993). Nevertheless, further
structural characterizations are necessary to understand the small satellite peaks
observed in the LC phase, which indicate in-plane correlations.
Results and Discussion
52
Lac 2
Figure 4-13 (upper graph) shows the DSC scan of Lac 2. In comparison to that of
Lac 1, the main transition peak was broadened and the Tt was reduced to Tt = 50
°C. The phase transition enthalpy was also clearly reduced to DH = 9.2 kcal/mol. A
broad enthalpic peak was still observed at around Tp = 40 °C, does not however
correspond to any changes in lamellar spacing or chain packing. The small angle
X-ray scattering data at T = 20, 40, 60, and 80 °C (Figure 4-13, lower graph)
showed periodic lamellar structures. Across Tt = 50 ºC, the low angle spacing was
changed from 87 Å at T < Tt to 78 Å at T > Tt, respectively. The reproducibility of
the SAXS data was checked by the measurement of a different sample at LURE.
Here, the wide angle patterns at T < Tt (T = 20 °C) can be characterized with only
Table 1 The measured phase transition temperatures (Tt, Tp), phase transition enthalpy (∆H) as determined by DSC, and the low- and wide-angle spacing (dSAXS, dWAXS) with their identified phases for Lac 1 -3 are summarized.
Lac 1 Lac 2 Lac 3
Tt / Tp [°C] 74 / 60 50 / 40 --* DSC
∆H[kcal/mol] 30 9.2 --
dSAXS [Å] 68 -- 108 Lc, Lc�
dWAXS [Å] 3.76, 4.45,
7.50#
-- 4.19, 4.46,
7.61#
dSAXS [Å] -- 87 -- Lβ�
dWAXS [Å] -- 4.17 --
dSAXS [Å] 60 78 --
Phase
Lα
dWAXS [Å] 4.57 4.45 --
* no phase transition was observed until T=80°.
# The diffraction peak corresponds to the head group correlation.
Results and Discussion
53
one sharp scattering peak at 4.17 Å, which corresponds to the gel (Lß) phase. The
absence of a pronounced shoulder denotes that the alkyl chains have nearly no tilt
(Hinz, Kuttenreich et al. 1991; Köberl, Hinz et al. 1998). No correlation between
the lactose head groups could be seen, indicating that the head groups are
already hydrated in this phase. At T > Tt (T = 80 °C), a broad band at about 4.45 Å
could be observed, which is consistent with the fluid Lα phase without any head
group correlation. The number of the equi-distanced peaks in the small angle
scattering were smaller at T > Tt.
Here can be concluded that the Lac 2 lamellar has a transition between the gel
Lac 2
20 40 60 80
0
2000
4000
6000
T [ ºC]
0.60.50.40.30.20.1
20 C
40 C
60 C
80 CSAXS WAXS
q [Å –1]
Figure 4-13 (Upper) Heat capacity trace of the Lac 2 dispersion (1 mg/mL), showing a broadened transition peak at Tt = 50 ûC and a distinct pre-transition at around T = 40 ûC. The phase transition enthalpy was also clearly reduced to ∆H = 9.2 kcal/mol. (Lower) SAXS diffraction patterns of the lamellar dispersion of Lac 2 at T = 20, 40, 60, and 80 ºC (left). The lamellar spacing showed a transition between 40 ºC (dSAXS = 87 Å) and 60 ºC (dSAXS = 78 Å). WAXS peaks suggested the transition between the gel phase and the fluid phase.
Results and Discussion
54
phase and the fluid Lα phase caused by a shift of the hydrophobic/hydrophilic
balance between the head groups and the alkyl chains. This shift towards the
hydrophilic side reduces the cooperativity between the alkyl chains, resulting in the
decrease in the transition temperature and the phase transition enthalpy. The
strongly crystallized alkyl chain packing modulates to the gel phase, which allows
for the hydration of the head groups.
The change in the Gibbs free energy between the two phases can be expressed
as
STHG ∆−∆=∆ . Eq. 4.9
At the phase transition temperature Tt, DG = 0 and therefore
tTHS /∆=∆ . Eq. 4.10
The transition enthalpy and temperature could be measured experimentally; the
entropy can be calculated by Eq. 4.10. This leads to a change in phase transition
entropy:
molKkcalSSS LacLac /59)( 12 −=∆−∆=∆∆ Eq. 4.11
The decrease in transition entropy from Lac 1 to Lac 2 agrees well with the
morphology suggested by X-ray diffraction. Below the phase transition
temperature, Lac 1 is in the highly ordered crystalline LC phase, while Lac 2 takes
the gel (Lß) phase due to the hydration of the head groups. Thus, the higher
degree of order in the crystalline phase can be related directly to the difference in
the phase transition entropy.
Lac3
In contrast to Lac 1 and Lac 2, the phase behaviour was significantly changed
when the number of lactose units was increased to N = 3. The DSC traces (upper
graph Figure 4-14) showed much less remarkable peaks until T = 80 ºC, which is
close to the highest operating temperature for aqueous dispersion. The SAXS data
(lower graph Figure 4-14) exhibited more than 10 equi-distanced peaks,
suggesting a highly ordered lamellar structure, which remained constant at 108 Å
Results and Discussion
55
between 20 ºC and 80 ºC. For the sample with a concentration of 50 wt%, the
SAXS data measured at LURE confirmed the reproducibility. The wide angle
scattering patterns (lower graph Figure 4-14) suggested no transition, exhibiting
three sharp scattering peaks at 4.19, 4.46, and 7.61 Å. These results clearly
indicate that the Lac 3 lamellar has no chain melting. The sharp peak at 7.61 Å
can be attributed to the strong head group correlation between the dehydrated
head groups. Judging from the peaks at 4.19 and 4.46 Å, dihexadecyl chains of
Lac 3 take a highly packed crystalline-like phase with a slight tilt or defects.
Moreover, it is also confirmed that the very weak enthalpic peaks around 25 ºC
and 55 ºC do not induce any morphological transition.
20 40 60 80
Lac 3
0
2000
4000
-2000
T [ ºC]
WAXS
2.01.81.61.41.21.00.8
80 ºC
20 ºC
4.19 Å4.46 Å
7.61 Å
q [Å–1]
SAXS
80 ºC
20 ºC
0 0.1 0.2 0.3 0.4 0.5 0.6
60 ºC
40 ºC
Figure 4-14 (Upper) DSC trace of the Lac 3 dispersion (1 mg/mL), showing no evidential endothermic peaks. (Lower) SAXS diffraction patterns of the Lac 3 lamellar dispersion at T = 20, 40, 60, and 80 ºC (lower left). The lamellar spacing showed no transition at all measurement conditions, dSAXS = 108 Å. WAXS peaks suggested that the Lac 3 lamellar takes crystalline-like phase and no chain melting takes place.
Results and Discussion
56
The very strong correlation between the hexasaccharide head groups obviously
forced the alkyl chain to take the tight, crystalline-like packing, which is different
from the ideal hexagonal lattice. Since the attractive interaction between the head
groups is strong, hydration can no longer take place.
�Hydrophobic� appearance of linear oligo- and polysaccharides has been well
known for cellooligosaccharides and cellulose. For example, Sano et al. reported
that cellooligosaccharides are mono-molecularly soluble in water when the
monosaccharide unit number was N = 1 ~ 4, whereas they can only be dissolved
in an aggregate state when N = 5 (Sano, Sasaki et al. 1991). Hato et al. reported a
similar phase for a lipid with two dodecanoyl chains and cellooligosaccharides with
N = 5, to which they gave the name �hydrated crystal� (Hato and Minamikawa
1996). But, the interpretation of this phase behaviour remained difficult. Indeed,
cellulose is insoluble in most solvents as well as in water (Brandrup and Immergut
1975). It has recently been shown, that the water uptake ability of the highly
ordered cellulose films is obviously poorer (Rehfeldt) compared to that of dextran
films (Mathe, Albersdorfer et al. 1999). We tentatively understand this LC� phase of
Lac 3 in terms of a �frozen� bilayer, which can appear either at very low
temperature conditions or at very high surface pressures (Lipowsky 1991;
Sackmann 1996). The WAXS peak positions can be related to the chain tilting, in-
plane defects, or the buckling induced by the strong head group correlation.
Electron density Profiles of Some Representative Phases
Structural analyses of several representative phases were attempted by
reconstruction of the electron density profiles (Harper, Mannock et al. 2001). The
measured SAXS data were fitted with Gaussians after subtraction of background
scattering. A Lorentz correction was applied by multiplying each peak intensity
(peak area) with its corresponding wave vector q (Warren 1969). Finally, the
square root of the corrected peak intensity was used to determine the constant
form factor F of each respective reflection. The electron density profile relative to
the constant electron density profile of water was calculated by the Fourier
synthesis according to Eq. 3.21. For centrosymmetric crystals such as lamellar
stacks of lipid bilayers, the electron density can be presented as a Fourier series
Results and Discussion
57
of cosines, therefore, the unknown phases are either 0 º (+) or 180 º (−). In the
following consideration, the origin was set to the center of the methyl dip of the
hydrophobic chains by fixing the phase of the first order reflection to �−�. All peak-
fittings and further calculations were carried out with the software package Origin
5.0 (Microcal Software, Northampton, U.S.A.).
Firstly, the SAXS data of Lac 1 at 80 °C and Lac 2 at 60 °C (Lα phase) were
analyzed. Each, four strong reflections h = 1, 2, 3, 4 of Lac 1 and h = 1, 2, 3, 6 of
Lac 2 were considered for the Fourier synthesis. Out of the possible 24 = 16
combinations, we chose 8 combinations that were centred in the middle of the
bilayers �� � � �, � � � +, � � + �, � � + +, � + � �, � + � +, � + + �, � + + +�,
corresponding to the terminal methyl dip (�). The most plausible phasing �� � + ��
shows a good similarity in the hydrocarbon chains region to the very well studied
Lα phase of dipalmitoylphosphatidylethanolamine (DPPE) (Pabst, Rappolt et al.
2000), and displays the appropriate head group size: about 10 Å for Lac 1 and 20
Å for Lac 2, respectively. All the remaining combinations lead to inappropriate
structural features, such as too large hydrocarbon core, missing methyl dip, or too
small head group size. By assuming that the maximum of each electron density
profile in Figure 4-15 (left) displays the midpoint of head groups, thickness of the
alkyl chains dal can be estimated to be 15 - 17 Å, for both Lac 1 and Lac 2. This is
in good agreement with the corresponding value reported for DPPE of 15.4 Å at 74
°C. From the obtained dal value, the thickness of the water layer between two
bilayers was calculated to be 6 - 8 Å. SAXS diffraction pattern of the crystalline-like
phase of Lac 3 (at 20 °C) displays 10 diffraction orders (Figure 4-15 right), which
results in 29 = 512
Results and Discussion
58
different possible phase combinations. The simple approach to choose the most
reasonable matching from all possible results obviously fails in this case.
Therefore, we have developed a simple three strip model for the Lc phase of Lac 3
(Figure 4-15 right (bottom)), based on the lactose head group, the hydrocarbon,
and the mid-plane region. Here, the water layer was not taken into account
because strong head group correlation in the crystalline phase of Lac 1 and Lac 3
suggested that there should be no bulk water between the bilayers. The electron
density of the head group was estimated from the density of lactose of 1.525 g/cm3
and its molar mass of 342.0 g to be about 0.48 e/Å3.
Table 2 Summary of the Fourier coefficients Fh, which have been used to determine the electron density maps of Figure 4-15.
h Fh
(Lac 1 at 80°C)
Fh
(Lac 2 at 60°C)
Fh
(Lac 1 at 20°C)
Fh
(Lac 3 at 20°C)
1 -1 -1 -1 -1
2 +0.22 +0.58 +0.11 -0.28
3 -0.18 -0.15 -0.07 -0.03
4 -0.10 ---- -0.17 +0.04
5 --- ---- --- -0.05
6 --- -0.04 -0.10 -0.08
7 --- --- --- -0.05
8 --- --- --- +0.03
9 --- --- --- -0.04
10 --- --- --- -0.03
Results and Discussion
59
The electron density of the hydrocarbon region, 0.30 e/Å3, and the terminal
methyls, of 0.16 e/Å3, were taken from the work of Harper et al. (Harper, Mannock
et al. 2001). Width of the head group region was set to 30 Å by assuming a
cylindrical conformation, while that of the methyl trough was assumed to be 8 Å
(Wiener, Suter et al. 1989). The phasing that results in the electron density plot
with the smallest mean absolute deviation to the simple three-strip model is given
in Table 2. The final electron density profile is superimposed to the model shown
in Figure 4-15 right (bottom). It is noteworthy that the given resolution enables one
to distinguish each lactose unit at the positions of about z = +/− 29, +/− 40 and
+/− 52 Å. SAXS data of Lac 1 at 20 °C, corresponding to crystalline phase, was
analyzed by taking the reflections h = 1, 2, 3, 4, 6 into account. Among the 16
possible solutions 4 reasonable candidates were found �� � � � �, � + � � �, � � +
� �, � + + � �� to be consistent with typical lipid bilayer features. Here, we chose ��
+ � � �� as the final solution (Figure 4-15, right, upper) since the corresponding
electron density profile shows the best similarity in the hydrocarbon chains region
with that of Lac 3 at 20 °C. The head group centre at z = +/− 29 Å almost coincides
with the first lactose position of Lac 3, and the alkyl chain length dal is about 24 Å in
both crystalline phases. Here we refrained from determining the electron density
profile of Lac 2 at 20 °C (gel phase), because the sub peak between the h = 2 and
-40 -20 0 20 40
ρ(z)
[arb
. uni
ts]
z [Å]-60 -40 -20 0 20 40 60
ρ(z)
[arb
. uni
ts]
z [Å]
Figure 4-15 (Left) Electron density profiles calculated from SAXS diffraction patterns of Lac 1
at 80 ºC (top), and that of Lac 2 at 60 ºC (bottom). Under these conditions, the glycolipids are in fluid Lα phase. (Right) Electron density profile of crystalline-like Lac 1 at 20 ºC calculated from h = 1, 2, 3, 4, 6 (top) and that of Lac 3 at 20 ºC obtained from a simple three-strip model (bottom).
Results and Discussion
60
3 could not be explained.
4.1.5 Summary
Phase behaviour, forces and morphology of three different glycolipids, Lac 1 -3,
with linear head group conformation were studied in monomolecular films and
dispersions. When the length of the head group is increased, the phase transition
pressure rises and the monolayer swelling behaviour (�out of plane elasticity�)
becomes more polymer-like. This is understood in terms of stronger sterical
interactions between head groups due to a higher number of monomer
(carbohydrate units) unit. Quite different, the �in plane viscoelasticity� does not
change continuously with the length of the head group. For short head groups the
hydrophobic chain-chain interactions are dominant and addition of carbohydrate
dimers disturbs these interactions causing a decrease in viscoelasticity with the
monolayer being predominately viscous. For 3 carbohydrate dimers (Lac 3)
however, the head group interactions dominate the system resulting in an increase
in G� and G�� and a rather elastic monomolecular film. In addition, for Lac 3 a
coupling between thermodynamic (1st order) and rheological transition by the
formation of physical network of hydrogen bonds has been observed. Both DSC
and especially X-Ray scattering experiments are in excellent agreement with the
molecular picture outlined above, suggesting the hydrophobic/hydrophilic balance
between chains and head groups being the driving mechanism for the �in plane
viscoelasticity� of the three glycolipid monolayers studied in this section. For short
head groups the strong chain chain correlations found in SAXS combined with the
DSC profile propose the chains being in a crystalline phase, which even causes
the head groups to dehydrate (concluded from the sharp head group correlation
peak). As the head group length is increased their correlation peak disappears
caused by the hydration of the sugar moiety. This in turn reduces the strong
correlations between the hydrophobic chains causing a shift from crystalline to gel
phase. The DSC and X-ray scattering experiments of Lac 3 strongly suggest the
idea of a �frozen� bilayer. No clear phase transition could be observed. The WAXS
data propose the head groups being dehydrated, which supports the picture of a
strong network of hydrogen bonds. This behavior of a type of �critical length� of the
Results and Discussion
61
head groups in order to shift the balance from hydrophobic to hydrophilic
interactions is similar to the results by Sano (Sano, Sasaki et al. 1991) who found
that cellooligosaccharides are soluble in water up to a number of 5 monomers.
Above 5 monomers, these molecules form aggregates. All this shows the
enormous importance of understanding the interactions between carbohydrates
(especially hydrogen bonding) in order to correctly judge their role for membrane
mechanic and stability.
4.2 Glycolipids with Branched or Bent Head Group Conformation (Lewis X, Gentiobiose)
The experiments presented in the last section can explained to a significant
degree by the formation of hydrogen bonds between neighbouring carbohydrates.
This formation is strongly influenced by the sterical alignment of these sugars.
Therefore, two glycolipids one, with branched and one with a bent carbohydrate
head group, as opposed to linear (cylindrical), were studied. The first one,
gentiobiose lipid, is believed to be involved in the toxicity of lipoteichoic acid (LTA)
in gram positive bacteria, similar to the well known lipid A anchor of the
Results and Discussion
62
lipopolysaccharide (LPS) layer in gram negative bacterias (Morath, Stadelmaier et
al. 2002). Chemically, it is almost identical to Lac 1 with the only difference being a
junction between the glucose and galactose (compare Figure 3-13 and Figure
3-14) causing a slight bend. The second one, Lewis X lipid, partakes in the
adhesion process of leucocytes in cell rolling events (Figure 3-15).
4.2.1 Monomolecular Films of Gentiobiose Lipids.
Langmuir isotherms
To characterize the phase behaviour of gentiobiose lipid films Langmuir isotherms
were taken at different temperatures (Figure 4-16). Tentatively, the same
behaviour was found as for the three Lac lipids, although the phase transition
pressure is about 5 mN/m higher at otherwise identical conditions. Following the
concept outlined in section 3.1.1 a tricritical pressure around 18 mN/m at an area
per molecule of about 64 Å2 can be calculated. The latent heat decreases linearly
from approximately 80 kJ/mol at T= 283 K to 40 kJ/mol at 302 K, which is slightly
lower than that for Lac 1 due to the decreased cooperativity of hydrophobic chain
� chain interactions.
pC=18.0 mN/m
Figure 4-16 Langmuir isotherms of the Gentiobiose Lipid monolayers at different temperatures. The thermodynamic evaluation was done in the way described in section 2.1.1.
Results and Discussion
63
ISR Experiments
After the linear viscoelastic regime was found to be around 100 µm, the amplitude
was set and fixed at that number. The frequency was fixed to 0.92 rad/s, where no
strain rate induced structural changes were found in the frequency sweeps done
prior to every experiment. In Figure 4-17 the rheological numbers of a gentiobiose
lipid monolayer are shown.
A
B
C
AB
C
Figure 4-17 Dynamic moduli of a gentiobiose monolayer on H2O as a function of area per molecule. Correspondence between dynamic moduli and the Langmuir isotherm (given in the inset) is indicated. The strain amplitude was kept in the linear response regime (100 µm) throughout the experiments. T = 20�C, f = 0,92 rad/s.
Results and Discussion
64
As the film was compressed below the liquid expanded � liquid condensed phase
transition (marker A in Figure 4-17), the viscoelasticity shows a drastic increase,
followed by a continuous one. This is similar to the behavior found for Lac 2 in the
last section (Figure 4-7). At an area of 41Å2 , which corresponds to a surface
pressure of 50 mN/m, the film reached its collapse pressure with G�=6.6 mN/m and
G��=7.4 mN/m. Over the whole range of surface pressure the film behaved more
viscous than elastic (G�>G��), although the ratio between G�� and G� decreased
from 6 (marker A) to a factor of almost 1 at marker C. The enormous decrease in
viscoelasticity, with respect to the Lac 1 monolayer, is again explained in terms of
hydrophobic/hydrophilic balance. The branched conformation of the gentiobiose
lipid causes a better hydration of their head groups which in turn decreases the
hydrophobic chain-chain interactions. The balance is shifted towards the
hydrophilic head group. However, as was shown for Lac 3, the viscoelasticity of
these monomolecular films of glycolipids do not only increase due to stronger
hydrophobic interactions or weaker hydrophilic disturbance, but also due to an
enforced network of hydrogen bonds. To manipulate these hydrogen bonds
between the head group for identical molecules, we replaced the subphase water
by D2O. In Figure 4-18 the experiments on this subphase are presented. Even
A
B
C
A
B
C
Figure 4-18 Dynamic moduli of a gentiobiose monolayer on D2O as a function of area per molecule measured at a strain amplitude of 100 µm. Correspondence between dynamic moduli and the Langmuir isotherm is indicated (given in the inset). T = 20°C, f = 0,92 rad/s.
Results and Discussion
65
though no change in the isotherms was observed, a clear shift in G� and G�� can be
seen. Obviously, the D2O causes a stronger network of hydrogen bonds between
the polar head groups of the gentiobiose lipid. This effect is in accordance with the
stronger dipole moment of D2O and was also found on thin polymer films
(Naumann, Brooks et al. 1999). Again it was nicely demonstrated that the
viscoelastic behavior of glycolipid films is determined by the balance between
hydrophilic and hydrophobic interactions. Furthermore, from the comparison
between isotherm and ISR experiments it can be stated that new insight into the
viscoelastic properties of glycolipid monolayers was gained from these in plane
rheology experiments.
X-Ray and DSC
The interpretation of our experiments is strongly supported by DSC and x-ray
scattering experiments (HASY Lab, Hamburg, Germany), which showed prominent
indications of a Lβ to Lα phase transition of hydrated bilayers (or lamellar stacks of
bilayers), i.e. no dehydrated crystalline phase with strong head group correlations,
as in the case of Lac 1, was observed. This is obvious from the shift in lamellar
spacing from 6.8 nm to 6.1 nm (Figure 4-20 left), as followed by SAXS as well as
10 20 30 40 50 60 70 80
0
2000
4000
6000
8000
c p[cal
/mol
°C]
Temperature [°C]
Figure 4-19 DSC data of Gentiobiose lipid vesicles. The phase transition enthalpy was calculated to be about 8 kcal/mol, similar to the Lβ to Lα transition of DPPC or Lac 2.
Results and Discussion
66
from the WAXS peak at 4.2 Å (Figure 4-20 right), which clearly correlates with the
melting transition observed in the DSC experiment (Figure 4-19). The change in
the SAXS peak position between 45 and 50 °C can be partially attributed to the
thermal expansion coefficient of the lipids, which were calculated to be
approximately -2.2*10-3 1/K, which is comparable to the one for DPPC (-3*10-3
1/K) (Sackmann 1996). This shows once more, that the hydrophilic interactions for
these lipids are not strong enough to dominate the physical behaviour of the
system. Also, already during preparation of the dispersions a very good solubility
of the lipid vesicles in water was obtained, which is in accordance with the smooth
change in the cp trace and results from the good hydration of the head groups.
Furthermore, no hysteresis for subsequent scans was detected.
0.10 0.15 0.20 0.25
400
8001200
6.1 Å
6,8 Å 40 °C
s [nm-1]
0500
10001500
44 °C
500100015002000
I [a
rb. u
nits
]
45 °C
500100015002000
50 °C
1.5 2.0 2.5 3.0
0
1000
2000
q [Å-1]
4,2 Å
20 °C
0
1000
2000
I [a
rb. u
nits
]
80 °C
Figure 4-20 SAXS (left) and WAXS (right) data of lamellar dispersions (50 wt% water) of
Gentiobiose lipid bilayers. Both indicate a phase transition. The WAXS peak (right) at 4.2 Å suggests a Lβ phase.
4.2.2 Monomolecular Films of Lewis X Lipids.
In this subsection, the physical behaviour of thin films of Lewis X lipids will be
discussed. The Lewis X head group is known to take part in the cell rolling process
during leukocyte adhesion. Concerning the conformation of the head group, it�s
important to notice that they take a branched rather than linear conformation.
Langmuir Isotherms
Langmuir isotherms taken at four different temperatures are shown in Figure 4-21.
Results and Discussion
67
Again (cf. Figure 4-1 and Figure 4-2), a continuous (linear) increase in phase
transition pressure pk and decrease in phase transition enthalpy (calculated
following Eq. 4.1) can be observed. For 17°C (290 K) the heat of transition is
approx 15 kJ/mol, which is only about 50% of that observed for Lac 2 and Lac 3
(Figure 4-3). Following the same arguments as above, this can be attributed to the
decrease in cooperativity between the hydrophobic alkyl chains caused by the
disturbing influence of the bulky hydrophilic head group. Also the minimal area of
approximately 65 Å2 clearly demonstrates the strong influence of the head groups
on the isotherms. Since the hydrophobic backbone has a minimal area of
approximately 37 Å2 (cf. Figure 4-1) and the cross section area of a sugar
molecule is about 20 Å2 , the minimal area of approximately 60 Å2 suggest a non
perpendicular alignment of the trisaccharide Lewis X with respect to the air/water
interface. This conformation is supported by recent x-ray crystallography
experiments by Üerez and co-workers (Perez, Mouhous-Riou et al. 1996), who
found a rod like conformation of the trisaccharide head group Lewis X in a highly
hydrated environment.
ISR Experiments
As in all the other ISR experiments, the range of linear viscoelasticity first was
determined. From this the frequency was set to 1 rad and the amplitude to 50 µm.
Area per Molecule [Ų]
Figure 4-21 Isotherms of Lewis X at four different temperatures. Heat of transition were extracted as described in Figure 4-1
Results and Discussion
68
Completely different from the rheology of Lac 1 -3 and gentiobiose lipid, the
viscoelasticity of Lewis x lipid showed a transient changes in G� and G�� (Figure
4-22). As the film was compressed below 70 Å2 , G�� (the viscosity) increased until
it reached a maximum of 0.4 mN/m at approx. 64 Å2. Further compression
decreased the viscoelasticity of the monolayer down to almost 0 mN/m at 58 Å2. If
the film was even further compressed below this area the viscoelasticity again
started to rise, until the film collapsed. Subsequent expansion and compression of
the film revealed the numbers for G� and G�� with an error bar of about 10 %,
accounting for the reversibility of this transient change. Obviously, a more complex
mechanism, as believed for Lac 1 - 3 and gentiobiose lipid, determines the
surprising rheological behavior of Lewis X lipid monolayers. The behavior found
here might be explained by an isotropic � nematic transition as it is known from
liquid crystals. For high areas per molecule, the system was in an isotropic phase.
Compressing the film to lower areas per molecule eventually lead to a strong
increase in molecular interaction between neighboring molecules and hence rise in
the viscoelastic constants. This is visualized in Figure 4-23. Decreasing the area
per molecule further lead to a reorientation of the rods which caused a drastic
reduction in molecular interaction (higher rotational entropy). The following
continuous increase in viscoelasticity until the collapse pressure is reached can be
explained by an increase in molecular interaction due to an increase in lateral
density. The rod like conformation parallel to the air/water interface and therefore
an anisotropic two dimensional layer of molecules is supported by our isotherms
as well as by x-ray crystallographic experiments ((Perez, Mouhous-Riou et al.
1996). The theory from Maffetone and coworkers predicts a transition from the
isotropic to the nematic state for cL2 = 2.1 � 2.4, where c is the surface
concentration and L the length of the rod (Maffettone, Grosso et al. 1996). Taking
c = 1/64 Å2 from the isotherm and L = 12 Å from the mentioned x-ray experiments,
we end up at cL2 = 2.3 � 2.5, i.e. in excellent agreement with the theory.
Therefore, the transition can be considered as an isotropic nematic transition of a
two dimensional liquid crystal.
Results and Discussion
69
Area per Molecule [Ų]
Figure 4-22 G� and G�� of Lewis X as a function of area per molecule. Clearly, the transient change in G� and G�� can be seen.
It is very important to note that no indication for such a transition was expected
from the Langmuir isotherm (Figure 4-21), which again clearly shows that the ISR
gives new insight in the physical properties monomolecular lipid films, which can
not be extracted from Langmuir isotherms.
Isotropic (semi-dilute)(Increase of G�� ~ 1/Area)
Top View:
Nematic(Decrease of G��)
n
l
Side Top
L~12Å
Figure 4-23 Sketch of the projection of a Lewis X molecule on the surface (upper) and an isotropic nematic transition (lower).
Results and Discussion
70
4.2.3 Summary
In this section the viscoelasticity of two synthesized glycolipids were studied by
ISR. Prior to all experiments, the linearity of the stress-strain relationship was
assured by amplitude sweeps. In addition, the absence of frequency induced
changes in G� and G�� was certified around 0.15 Hz.
The gentiobiose lipid, which has identical chemical composition as Lac 1 but
differs only in head group conformation (β 1, 4 junction), caused a drastic
fluidization of the monolayer. Again, this was because of disturbed chain � chain
interactions by the hydration of the head groups, which resulted in a shift of the
hydrophobic/hydrophilic balance. This is also evident from the x-ray scattering
experiments. The gentiobiose lipid showed a gel Lβ to fluid Lα phase transition,
while for Lac 1 a crystalline Lc to fluid Lα phase was observed. Furthermore, an
increase in viscoelasticity could be achieved by exchanging the subphase with
D2O, clearly showing the importance of hydrogen bonds and
hydrophilic/hydrophobic balance in viscoelasticity studies of membranes. The
branched tetrasaccharide Lewis X head group revealed a fluid crystalline behavior.
At a surface pressure of approximately 25 mN/m an isotropic to nematic phase
transition took place, which manifested as a transient change in the viscosity. This
was a result of the anisotropy of the Lewis X molecules, when aligned as little rods
of about 12 Å parallel to the air/water interface. Above the transition, i.e. for high
surface pressures, when all rods are aligned in the same direction and tightly
packed together a continuous increase in viscosity was observed owing simply to
the reduction in area per molecule.
It should again be emphasized that neither the viscoelastic behavior of gentiobiose
nor the curious transition in Lewis X lipid monolayers could have been predicted or
explained from Langmuir-isotherms. This is obvious, since i) the isotherms of
gentiobiose lipid on D2O and H2O are identical and ii) absolutely no indication of a
transition from isotropic to nematic can be extracted from the Lewis X lipid
isotherm. Clearly, we have gained new insights into the physics of 2D
monomolecular films of glycolipids.
Results and Discussion
71
4.3 Phase Behaviour of Fluorinated Lipids and Artificial Microdomains
Fluorinated lipids are promising candidates for a variety of applications, e.g. as
drug delivery systems. This is due to the high electronegativity of fluorine
Results and Discussion
72
(especially with respect to hydrogen) which causes strong dipole-dipole repulsion
when parallel oriented and because of its hydro- and lipophobicity. This builds the
basis for the design of small lipid domains when mixed with hydrocarbon chains.
Both in plane dipole � dipole interactions as well as microdomain formation will be
explained in this section.
4.3.1 Stripe-like Phase Formation in Fluorinated Lipid Monolayer (Schneider, Andelman et al. 2003)
The shapes and shape transitions of domains in lipid membranes, accompanied
by phase transitions, have been a subject of intensive research for many years.
This arises from the interest in the formation process of lipid microdomains (e.g.
rafts) in biomembranes as well as from fundamental questions about the self-
assembling or pattern formation of lipid microstructures. In the following section
the formation of a modulated stripe-like phase is discussed. It is shown for the first
time that this formation is based on strong dipole-dipole interactions and not on
reduced line tension as discussed previously (Keller, McConnell et al. 1986).
4.3.1.1 Experimental Observation
Figure 4-24 shows the fluorescence images taken at a mean molecular area of
90Ų per lipid molecule or 45 Ų per single chain. The thickness of the stripes
50 µm50 µm
Figure 4-24 Fluorescence pictures of fluorinated lipid monolayer taken at 45 A² per lipid chain, where the film is in the gas-liquid coexistence phase. The fluorescence probe concentration was 0.1% and the line thickness measured from these pictures varies between 2 � 8 µm.
Results and Discussion
73
obtained from different images ranges between 2 � 8 µm. In the Langmuir
isotherm (Figure 4-26) this means that the molecular area is close to the onset of
the surface pressure increase, suggesting that it is close to the first order phase
transition from gas to liquid-expanded. No fluorescence signal could be detected
for mean molecular areas above 100 Ų as would be expected for the gas phase,
while homogenous fluorescence images could be obtained for mean molecular
areas below 80 Ų, which is typical for liquid-expanded phases. Therefore, the
picture in Figure 4-24 corresponds to the coexistence of liquid-expanded and gas
phases. Surprisingly, the shape of the bright (i.e. fluid domains), appeared to be
stripe-like rather than circular, as observed for other non fluorinated lipids (Figure
4-25) (Möhwald 1995). It is also of interest to note that the domains in Figure 4-24
do not merge with each other and have rather sharp domain walls. To eliminate
kinetic effects arising from the finite compression speed, the monolayer was kept
at constant area (90 Ų per molecule) for more then 30 min. Subsequent
fluorescence images taken at 5 min intervals, showed no significant changes in
shape or thickness of the domains. We conclude that the observed structures are
thermodynamically stable. This is a necessary knowledge for the application of
any theoretical assumptions on the nature of these interesting domain shapes.
The shape and thickness of such domains is determined by two opposing forces
(McConnell 1991). One of these two forces is the line tension, which reflects the
energy needed to create a domain boundary between two phases. This energy is
proportional to the length of the line surrounding the domain and favors circular
shapes. The second force arises from the dipole moment of the lipid which tends
to elongate the domains due to their electrostatic interaction.
Results and Discussion
74
The effective molecular dipole moment in the Langmuir monolayer can analytically
be estimated by surface potential measurements (Brockman 1994) using the
Helmholz equation,
AV
εµ= Eq. 4.12
where µ is the molecular dipole moment, A the area per moment and ε the local
dielectric constant. In Figure 4-26 the surface potential of the fluorinated lipid
studied is shown as a function of the area per fluorinated chain. This figure also
50 µm
Figure 4-25 Fluorescence pictures of non fluorinated lipids taken at 30 A² per lipid chain, where the film is in the gas-liquid coexistence phase. The fluorescence probe concentration was 0.1%. No stripe-like domains can be observed.
Area per Chain [ Ų]
Figure 4-26 Surface potential and pressure-area isotherms of the fluorinated lipid monolayer. The onset of the transition to the condensed phase correlates with the abrupt decrease in surface potential. An average surface potential in the gas-liquid coexistence region around � 430 mV (Ā = 45 A²) was found.
Results and Discussion
75
includes the simultaneously measured pressure-area isotherm (area per chain
instead of molecule). For relatively large areas per lipid chain the surface potential
remains relatively constant and stripe like domains as shown in Figure 4-24 can be
observed. Compression of the film eventually leads to a steep increase in surface
pressure, indicating the end of the gas-liquid expanded coexistence. This sudden
increase in surface pressure correlates with an abrupt decrease in surface
potential. This can be attributed to an effective increase in the number of dipoles
aligned perpendicular to the subphase as a result of the higher chain ordering.
Figure 4-26 demonstrates two remarkable features of the surface potential which
are in juxtaposition to those of ordinary fatty acids or phospholipids: i) the surface
potential is negative and ii) compression results in much higher absolute values for
small areas per chain. The potential of the fluorinated lipid monolayer at mean
molecular areas of 45 Ų per lipid chain, where the monolayer is in the gas � fluid
coexistence (Figure 4-24), ranged between -420 mV and -340 mV. Since the lipids
used in this study are not charged, the measured surface potential should be
dominated by the dipoles of the hydrophobic tails.
Three contributions are generally taken into account when examining the effective
molecular dipole moments in Langmuir monolayer (Brockman 1994): (i) the
contribution from the polar head group of the lipid, (ii) the influence of oriented (i.e.
polarized) water adjacent to the head group and (iii) contributions of asymmetric
chain termini. In the experimental system, the intermediate chain regions do not
contribute to the net molecular dipole, because the successive dipoles of the
groups along the chain cancel each other out as a result of their structural
symmetry. The contributions from (i) and (ii) are basically identical for fluorinated
and non fluorinated lipids (Smondyrev and Berkowitz 1999) therefore, the
difference in dipole potential arises from the fluorinated chain termini. Because the
surface potential of non fluorinated lipids in the gas or liquid-expanded phase is
close to 0 V (Vogel and Mobius 1988; Oliveira Jr, Taylor et al. 1992), the molecular
origin of the measured surface potential in the range of -420 mV to -340 mV can
be entirely attributed to the terminal �CF3 group. Thus, it is concluded that this
strong dipole moment determines the formation of stripe-like domains in the two
phase coexistence region.
Results and Discussion
76
The influence of line tension on the shape of stripe-like domains was studied by
adding a small amount of cholesterol which is known to reduce line tension
significantly (Weis and McConnell 1985). When 0.1 mol% of cholesterol was
added to the fluorinated lipids, the average thickness of stripe domains was
decreased to 1 � 2 µm (Figure 4-27). This can be understood as a result of the
above outlined picture of opposing forces and is described theoretically in the next
section.
50 µm
Figure 4-27 Fluorescence pictures taken at 45 A² per lipid chain with the film in the gas-liquid coexistence phase. The fluorescence probe concentration was 0.1% and in contrast to Figure 4-24 , the film contains 0.1 mol% cholesterol. The average line thickness found decreased with respect to Figure 4-24 down to 2 µm.
4.3.1.2 The Free Energy of a Patterned Field of Dipoles
A first-order phase transition between gas and liquid-expanded or liquid-expanded
and liquid-condensed can be induced in a Langmuir trough by decreasing the
mean molecular area of the film. The transition occurs as the attractive interaction
between molecules starts to dominate over the entropy of mixing. The first-order
coexistence curve in the temperature � density plane terminates with a critical
point (Albrecht, Gruler et al. 1978). Electrostatic interactions alter the above
mentioned gas-liquid expanded transition as a result of their tendency to stabilize
phases with modulated density (Andelman, Brochard et al. 1987).
Starting from dipole�dipole interactions the electrostatic free energy of a stripe�
like phase pattern following Keller (Keller, McConnell et al. 1986) and Andelman
Results and Discussion
77
(Andelman, Brochard et al. 1987) can be derived. A dipole µ at a distance r from
another dipole sees the field created by the dipole and its image dipole µ�. The
interaction energy between the two �real� dipoles located at a distance h above the
air/water interface (Figure 4-28) is therefore
)(2)(
003
2
εεεπεµ
+=
rrg . Eq. 4.13
Where ε, ε0 represent the local dielectric constants seen by the dipoles (water)
and the dielectric constant of air respectively. This dipolar interaction energy is
long range and repulsive, since all dipoles point in the same direction. Translating
the dipoles from just above (air) to just below (water) the interface decreases the
interaction energy by a factor of 6400 as a consequence of the high dielectric
constant of water, illustrating the significance of this constant. Since the above
discussed -CF3 dipoles are not immersed in water, large contributions to the
dipolar interaction energy can be expected, supporting the formation of modulated
phases. Close to the critical point it is convenient to remain within the framework of
the Landau expansion and to consider the contributions from electrostatics and
line tension as additional spatial varying terms leading to a Landau-Ginzburg
expansion. However, for lower temperatures where domains are assumed to have
AIR
WATER
h
WATER
AIR
Figure 4-28 Sketch of monomolecular film at the air/water interface. The dipoles at a distance h above the interface are presented as little arrows (lower sketch). They are not immersed in water but �see� the dielectric constant of air and neighboring chains.
Results and Discussion
78
sharp walls, as found in our experiments (Figure 4-24), a low temperature (with
respect to the critical point) calculation seems to be more convenient.
From Eq. 4.13 the interaction energy between one dipole and an infinite line of
dipoles at distance x can be calculated:
20
23
2/3220
24 2)()(
1)( x
aady
yxaU Line εεε
εµεεεεµ
+=
++= ∫
∞
∞−
.Eq. 4.14
Where a=(area per dipole)0.5 is a microscopic cutoff length (distance between
dipoles). The interaction between two lines of dipoles of length l is
20
22
22
)( xla
alUU LineLines εεε
εµ+
== . Eq. 4.15
In one strip of width DL, there are DL/ a lines (Figure 4-28b). This leads to an
electrostatic energy of one single strip
∑=
−
+=
aD
n
Lel na
na
DlaF/
10
22 1)(
2εεε
εµ . Eq. 4.16
The sum leads to the energy of one single strip of dipoles build of many lines of
dipoles. Following the same concept the interactions between the stripes can be
calculated. This eventually leads to
aa
DL
lFigure 4-28b Definition of one strip - One strip is made up of multiple lines of dipoles with
the width a and length l. Its total width DL can be extracted from the fluorescence pictures.
Results and Discussion
79
[ ]
−−−+=
axD
Dbxx
akTbF GLGLel π
πφφπ
φφπ
)sin(log)()1( 23
223 Eq. 4.17
where x = DL/D = DL / (DL+DG), A1=φ and
)( 00
23
εεεεµ
+=
kTb .
The first two terms in Eq. 4.17 represent the overall average contribution to the
electrostatic energy and are independent of the periodicity D. The third term is an
exact summation of the inter-stripe electrostatic interactions (Keller, McConnell et
al. 1986). The total free energy difference between the stripe and the homogenous
phase is therefore
DaxD
DkTbF GL
γπ
πφφπ
2sinlog)( 23
+
−−=∆ . Eq. 4.18
Minimizing Eq. 4.18 with respect to D
βπ
π exp)sin(
=
xaDEqui Eq. 4.19
where 1)(
223 +
−=
GLkTb φφπγβ
This is the equilibrium thickness of the stripes and can be compared to the above
observed patterns. Due to the exponential dependence of D on ε, the choice of
ε is critical. Using ε=2 (approx. dielectric constant of fluorine media) decreases D
by a factor of approximately 100. However, since the dielectric contributions from
fluorine are already taken into account by their dipole-dipole interactions, e the
dielectric constant of air surrounding the -CF3 dipoles were chosen. Using γ =
1.6*10-12 N [Benvegnu, 1992 #12] leads to a stripe thickness in the range of 1 µm
for the measured surface potential, which is in excellent agreement with the
numbers taken from the fluorescence pictures. Since the above theory is an
equilibrium theory, the observed patterns can be considered as equilibrium
Results and Discussion
80
patterns rather than intermediate states as previously observed and theoretically
described in many nucleation and growth processes. From Eq. 4.19 one would
also expect that a decrease in line tension γ leads to a decrease in line thickness
D. This is in agreement with the observation (Figure 4-27) that small traces of
cholesterol (<0.1% mol) caused a decrease in the microscopically observed line
thickness and thus further supports the idea that this are thermodynamically stable
structures.
Finally, the phase diagram shall be constructed from the thermodynamic potential.
From Eq. 4.18 and Eq. 4.19 follows
)exp())((3
βφφφφπ
−−−−=∆ LGakTbF . Eq. 4.20
This is the free energy of the stripe phase. The thermodynamic potential of the two
isotropic phases (gas, fluid) close to their equilibrium order parameter can be
approximated by parabolas
21 )(21
GasGasGas kG φφ −= −
21 )(21
FluidFluidFluid kG φφ −= − .
Eq. 4.21
Where k represents the compressibilities of the gas and the fluid phases,
respectively. In Figure 4-29 the free energy (upper graph) and the phase diagram
is shown. The double tangent construction yields the phase diagram (lower graph
in Figure 4-29). Regions for pure gas, fluid (or liquid) and stripe�like phases are
found, separated by two small regions of coexistence between modulated stripe-
like and isotropic phases are found.
Results and Discussion
81
G
T
η
η
G G+S LS+LS
Stripe Phase
Isotropic Phase(Fluid)Isotropic Phase
(Gas)
Figure 4-29 (Upper Graph) Thermodynamic potentials if isotropic and stripe phases as a function of surface concentration. Double tangent construction yields the coexistence region. (Lower Graph) Phase Diagram of gas (G), liquid (L) and stripe (S) phases. The coexistence regions are calculated from the upper graph.
4.3.2 Design of Artificial Glycolipid Microdomains by Fluorinated Lipids (Gege, Schneider, et al 2003)
In the following subsection the application of partially perfluorinated lipids with
functionalized saccharide based head groups is demonstrated.
4.3.2.1 Mixing Behaviour in Monolayers
Perfluorinated lipids are known to take helical conformation (Bunn and Howells
1954). This is because of the high space requirement of the CF2 groups as
compared to CH2 groups (CF2 ≈ 1.5 CH2), where the chains take a rather zig zag
(trans or gauche) conformation. Instead of increasing the bond angle between two
C Atoms to account for the higher VdW radius of Fluorine, the chain introduces a
twist, hence a helical conformation arises. The different space requirements can
be clearly resolved in the Langmuir isotherms (Figure 4-30), where the isotherms
of two lipids with exact same head group, but different tails, one of them partially
fluorinated, are shown.
Results and Discussion
82
This and the enormous difference in electronegativity between fluorine and
hydrogen (cf. chapter 3.3.1) is believed to make the perfluorinated chains not only
hydrophobic, but also lipophobic ((Shin, Collazo et al. 1993) and thus potential
candidates for the design of microdomains in lipid membranes. The fluorinated
lipids used in the present study are only partially fluorinated (Figure 3-16), i.e. only
their eight terminal CH2 and CH3 groups are replaced by CF2 and CF3 groups,
respectively. As illustrated in Figure 4-31, this should allow the remaining
hydrocarbon chain to stay in their disordered configuration (of course not with
respect to their diffusion properties) even in highly compressed films. Therefore
the fluorescence marker T-Red, which only emits light in the vicinity of unordered
chains stays visible even in high compressed fluorinated lipid monolayer ( Figure
4-32 right), whereas a film of hydrocarbon chain based lipids appears almost
completely black (Figure 4-32, left) (Möhwald 1995). This unique behaviour of
partially fluorinated lipid chains enables us to visualize fluorine lipid domains even
Figure 4-30 Langmuir isotherms of fluorinated and none fluorinated lipid monolayers.
Results and Discussion
83
for high compressed films. This is important, since only such films (highly
compressed) can reliable be transferred to solid supports and therefore used for
further studies.
The fact, that the partially fluorinated chains mix non ideal with hydrocarbon chains
can be concluded from Figure 4-33. Obviously, at least for high surface pressure
(25 mN/m), where the film is supposed to be transferred, the two lipids neither mix
ideally nor completely demix, since both should result in an isotherm in the middle
between the two pure systems for a 1/1 mixture ((Gaines 1966)).
Figure 4-31 Sketch of the helical structure of fluorinated chains. The alkyl chains stay fluid due to the higher area uptake of the fluorinated chains.
Results and Discussion
84
50 µm50 µm
Figure 4-32 (Left) Films of lipid monolayers with regular alkyl chains become dark at high lateral pressure. (Right) Perfluorinated lipid chains stay bright even at high compression.
In Figure 4-34 the corresponding fluorescence images for a 10/90 mixture of
fluorinated S-Lex lipid and DMPC are shown. At low surface pressures a fluid film
with liquid-condensed domains of DMPC can be seen, while at higher pressures a
few bright domains appear in an otherwise black lipid matrix. Since at high surface
Pure DMPC50 / 50 Slex-F/DMPCPure Slex-F
Figure 4-33 Langmuir isotherms of DMPC, fluorinated lipids and mixtures. The non ideal mixing behaviour can be concluded from the isotherm of the mixture (green).
Results and Discussion
85
pressures the dye is supposed to accumulate only in the disordered region of the
fluorinated lipids, this bright spots can be addressed to be domains of such lipids.
Thus it has been shown, that in fact small domains of a few micrometer with
functionalized head groups can be designed.
The next step in testing about the general application of fluorinated lipids for the
design of microdomains with functionalized head groups was to exclude the effect
of the head group structure on the overall mechanism of the formation process.
For that reason the mixing behaviour for two more head groups was studied. In
Figure 4-35 the domains for LexF/DMPC (10/90) and S-LexF/DMPC (10/90) are
compared. Even though somewhat different in size, the results are qualitatively the
same. Systematic variations of the concentration ratio of fluorinated and none
fluorinated chains (Figure 4-36) were done for mixtures between the lipid anchors
(-OH head group). Again even though not identical in shape and size, a
continuous increase of the bright domains with fluorinated anchor concentration
Slex-F/DMPC: 10/90
50 µm
50 µm
Figure 4-34 The Langmuir isotherm and fluorescence images of the mixed S-LexF/DMPC monolayer (molar ratio: 10/90). As described in the text, the fluorescence lipids (Texas Red DHPE) accumulated in the domains of perfluorinated lipids. At high surface pressure (Π > 20 mN), fluorescent clusters of S-SlexF could be observed in the dark matrix of DMPC. Total diameter 250 µm
Results and Discussion
86
could be observed. Thus, it can be concluded, that the formation of microdomains
is independent from the head group structure of the lipids and is therefore
obviously dominated by the mixing behaviour of the fluorinated lipid chains. It
A B
100 µm100 µm
Figure 4-35 Impact of the head group functions on the micro-domains: A) S-LexF/DMPC (10/90) monolayer at Π = 30 mN/m, and B) LexF/DMPC (10/90) monolayer at Π = 30 mN/m. Although the size of clusters appeared slightly different, the qualitative tendency strongly suggested that the hydrophobic mismatch between alkyl and F-alkyl chains plays a dominant role in formation of the micro-domains.
100% 90%
10%50%
50 µm
Figure 4-36 Impact of the mixing ratio on the micro-domains. For simplicity, F-alkyl lipid anchors were mixed with the alkyl lipid anchors. According to the increase in F-alkyl lipids from 10 % to 100 % a continuous growth of the fluorescent domains of F-alkyl lipids could be observed. Total diameter 250 µm.
Results and Discussion
87
should be noted that only the unique conformation of perfluorinated chains in
correspondence with the emission behaviour of the fluorescence probe used,
allowed for these important conclusions.
4.3.2.2 Mixing Behaviour in Bilayer (Vesicles)
To generalize and expand the concept of artificial microdomains giant lipid
vesicles of SlexF/DMPC mixtures (15/85) were prepared (appendix). In Figure
4-37 a typical fluorescence image of two vesicles (a small one inside a larger one)
is shown. The focus plane was set in the middle of the vesicle. All vesicles
appeared to be circular with a diameter between 5 � 50 µm. The picture was
chosen to illustrate the size distribution and similarities of domain appearance in a
single picture. The existence of fluorinated domains can be clearly seen. This
conclusion is drawn since, as shown in the last subsection, the fluorescence probe
accumulates in the fluorinated domains. The fact that the lateral pressure P
expected in lipid bilayer vesicles is above 25 mN/m, ((Möhwald 1995)) further
excludes that the bright spots seen, can be attributed to fluid like DMPC domains,
that reach the liquid-condensed phase around 5 � 10 mN/m. As a consequence
the probe must be diluted in the fluorinated domains. Even though the
fluorescence microscopy picture of Figure 4-37 clearly demonstrates the
occurrence of microdomains in giant lipid vesicles, a clear image of the shape and
size of the domains is lacking. Therefore, to gain the full 3D image of the vesicles,
scanning confocal fluorescence microscopy experiments were performed. In
Figure 4-38 a 3D reconstruction of a whole lipid vesicle from a stack of z-scans is
shown. Impressively the phase separation between the fluorinated SlexF and
DMPC domains as well as their size and shape on the vesicle surface is resolved.
Qualitatively the same formation of domains as in the monolayer studies is found.
Comparing the details of the domain shapes between the two systems seems
difficult, since the lateral pressure P in the bilayer is not know and the effect of
higher cooperativity in vesicles (additional cooperativity between the two
monolayer) as compared to monolayers is hard to account for.
Results and Discussion
88
10 µm
Figure 4-37 Fluorescence image of S-lexF lipid micro-domains (artificial lipid rafts) reconstituted in giant vesicles of DMPC (S-lexF/DMPC = 15/85).
5 µm
Figure 4-38 3D reconstruction of artificial rafts in a giant vesicle. It demonstrates that the F-alkyl lipids form self-assembled micro-domains in alkyl lipids both in monolayers at air/water interface as well as in vesicles with a single bilayer. Scale bar 5 µm.
Results and Discussion
89
4.3.3 Summary
The unique phase behaviour of fluorinated lipids and their mixing behaviour with
hydrocarbon chain based lipids has been intensively studied by fluorescence
microscopy and film balance experiments. For one component monolayers a stripe
like phase in thermodynamic equilibrium was observed and described on the base
of a theoretical model by D. Andelman (Andelman, Brochard et al. 1987). The
origin of this modulated phase was identified to be the strong dipole contribution of
the terminal CF3 groups. Apart from this, their distinctive mixing behavior was
shown to enable the controlled design of lipid microdomains not only in
monolayers, but also in lipid bilayers. Since the demixing is dominated by the lipid
chains, this holds in principle for every attached (especially functionalized) head
group. These results build the basis to study not only the effect of size and
concentration on cooperative binding events between membranes (ligand �
receptor), but also the influence of domain shape on such and other events (e.g.
enzymatic reactions).
Conclusions and Outlook
90
5 Conclusions and Outlook
In the present study the physical properties of lipids with and without fluorinated
tails and covalently attached carbohydrates (lactose oligomers, gentiobiose and
(sialyl) Lewis X) were systematically studied by applying six experimental
techniques. Thermodynamic and structural properties were obtained from film
balance, calorimetry (DSC), fluorescence microscopy, and x-ray scattering
experiments. Information about forces and viscoelastic properties within the plane
Conclusions and Outlook
91
of the membrane and perpendicular to the surface (�out of plane�) was acquired
with the use of interfacial stress rheometry (ISR) and ellipsometry under controlled
and variable humidity conditions. It was found, that �out of plane� and �in plane�
elasticity can behave differently. Namely, although the membrane becomes more
�soft� perpendicular to the surface when increasing the number of carbohydrates,
the lateral viscoelasticity increases when the number of monomers becomes N >
4, as a result of the non linear increase of hydrogen bonds between neighboring
carbohydrates with N. This is completely different from the behavior of
lipopolymers (e.g. PEG), in which the lateral viscoelasticity decreases dramatically
with increasing number of monomers. This information clearly demonstrates that
the synthetic glycolipids studied represent a more suitable model of the glycocalyx
to mimic and investigate the complex interplay of various physical forces in cell-
cell recognition processes. The ability of the glycolipids to form a physical network
of hydrogen bonds may for example help to protect the cell against the harsh
conditions frequently found on the apical side of epithelial cells. This network
formation was shown to be a cooperative process (i.e. at a certain number of
possible hydrogen bonding sites the network forms in an �all of a sudden� type
process) and can therefore be viewed as a molecular mechanism in order to
switch the macroscopic mechanical properties from viscous to elastic. In addition,
due to the fact that the elastic properties of the membrane are controlled or at
least modulated by the structure and length of the glycolipids, they can form a soft
cushion and so prevent non specific adhesion between cells or influence the
adhesion of vesicles, since adhesion is dependent on the elasticity of the
membrane. To further investigate the effects of carbohydrate-carbohydrate and
carbohydrate-protein interactions and move towards a more realistic physical
model of the glycocalyx the functional head group Lewis X was introduced.
Rheology studies of Lewis X displayed a fluid crystalline (anisotropic) behavior
with an isotropic to nematic transition, demonstrated by a maximum in surface
viscosity. This indicates the need to study the role of anisotropy or chirality in
membrane physics more intensively not only with respect to mechanical properties
as described here, but also for optical or electrical properties. An example is the
finding that smectic C fluid crystals display a piezoelectric-like effect (Brand and
Conclusions and Outlook
92
Pleiner 1984).
As a consequence of the thermodynamic analysis of glycolipid membranes it was
found that although the surface transition pressure and cooperativity change in
relation to the head group size, the monolayer maintains its overall stability
guaranteeing that the cell membrane will keep its integrity even under harsh
extracellular conditions.
The thermodynamic investigation of mixtures between fluorinated and non
fluorinated lipids revealed a strong demixing between the two compounds
independent of the hydrophilic head groups. Fluorescence microscopy proved the
existence of microdomains or artificial lipid rafts, a prerequisite for a variety of
biological processes, e.g. cell adhesion. Transfer of the monolayer, including the
designed functionalized domains (sialyl Lewis X), onto a solid support and its
subjection to a flow of CHO (Chinese Hamster Ovarial) cells, clearly confirmed the
impact of these domains on the cell adhesion process. The precise understanding
of the microdomain formation process gained throughout this work, will enable the
study of the effect of size and concentration as well as the influence of domain
shape on cooperative binding between membranes (ligand � receptor). The
additional fact that microdomains could be reconstituted in giant lipid vesicles
suggests an exciting potential for the design of new cell membrane models with
artificial rafts in 3D shells (�phantom cells�). This opens new possibilities for
studying cooperative interactions between membranes and proteins in the future.
Appendix
93
Appendix
A. Viscoelasticity of PEG-lipids
Here the viscoelasticity of Polyethylenglycol (PEG)-lipopolymers with different
head group length is briefly presented. As can be seen, there is a clear reduction
in viscoelasticity with increasing the number of polymers from N = 3 (figure A1) to
N = 9 (figure A2). This is in contradiction to the lactose lipids presented in chapter
4.1 (details there), in which the viscoelasticity increases with increase of sugar
Appendix
94
monomers from N=4 to N= 6 .The reason for this is the strong hydrogen bond
network sugars can form as discussed in detail in the chapter mentioned.
A
B
C
AB
C
Figure A1 G� and G�� of the PEG 3 lipid.
A
B
C
A
B
C
AB
C
A
B
C
Figure A2 G� and G�� of the PEG 6 (left) and PEG 9 (right) lipid.
B. Preparation of Giant Unilamellar Vesicles (GUVs)
GUVs were prepared using the electroswelling technique. The sample, dissolved
in chloroform, was spread on the electrodes and dried over night in a vacuum. A
Appendix
95
small amout (1/1000) of the fluorescence probe (Texas Red) was then added.
After complete evaporation of the solvent, an electric field of approximately 1V/mm
and 10 Hz was applied. To increase the weight of the vesicles the whole setup
(electrodes and chamber) was put into a sucrose solution (150 mM) during the
formation process. The operating temperature was set above the chain melting
temperature of the vesicles. After two hours the vesicles were put into a glucose
solution with slightly higher osmolarity (> 150mM) in order to avoid that the
vesicles burst. Lastly, small amounts of the sample were put on a cover slide and
observed under the microscope. The diameter of these vesicles ranged between 5
and 50 µm.
Literature
96
Literature
Ahrens, H., T. R. Bækmark, et al. (2000). �Hydrophilic/Hydrophobic Nanostripes in
Lipopolymer Monolayer.� ChemPhysChem 1: 101.
Alberts, B., D. Bray, et al. (1994). Molecular Biology of The Cell. New York &
London, Garland Publishing, Inc.
Albrecht, O., H. Gruler, et al. (1978). �Polymorphism of phospholipid monolayers.� Journal de Physique I 39(3): 301-13.
Literature
97
Alexander, S. (1977). J. Phys. II France 38: 983-987.
Andelman, D., F. Brochard, et al. (1987). �Phase transitions in Langmuir
monolayers of polar molecules.� Journal of Chemical Physics 86(6): 3673-81.
Benvegnu, D. J. and H. M. McConnell (1992). �Line tension between liquid domains in lipid monolayers.� Journal of Physical Chemistry 96(16): 6820-4.
Brand, H. R. and H. Pleiner (1984). �Macroscopic dynamics of chiral smectic C.� Journal de Physique I 45(3): 563-73.
Brandrup, J. and E. H. Immergut (1975). Polymer Handbook, Wiley Interscience.
Brockman, H. (1994). �Dipole potential of lipid membranes.� Chem Phys Lipids
73(1-2): 57-79.
Brooks, C. F., G. G. Fuller, et al. (1999). �An Interfacial Stress Rheometer To
Study Rheological Transitions in Monolayers at the Air-Water Interface.�
Langmuir 15: 2450 - 2459.
Bunn, C. W. and E. R. Howells (1954). �Structures of Molecules and Crystals of
Fluorocarbons.� Nature 174: 549 - 551.
Burack, W. R., Q. Yuan, et al. (1993). �Role of lateral phase separation in the modulation of phospholipase A2 activity.� Biochemistry 32(2): 583-9.
Caffrey, M. (1987). �Kinetics and mechanism of transitions involving the lamellar,
cubic, inverted hexagonal, and fluid isotropic phases of hydrated
monoacylglycerides monitored by time-resolved X-ray diffraction.�
Biochemistry 26: 6349 - 6363.
Cevc, G. (1993). Phospholipids Handbook. New York, Marcel Dekker.
Curatolo, W. (1987). �The physical properties of glycolipids.� Biochim Biophys Acta
906(2): 111-36.
Dietrich, C., L. A. Bagatolli, et al. (2001). �Lipid rafts reconstituted in model
membranes.� Biophys J 80(3): 1417-28.
Dietrich, C., Z. N. Volovyk, et al. (2001). �Partitioning of Thy-1, GM1, and cross-
Literature
98
linked phospholipid analogs into lipid rafts reconstituted in supported model membrane monolayers.� Proc Natl Acad Sci U S A 98(19): 10642-7.
Gabius, H. J. and S. Gabius (1997). Glycoscience, Chapmann & Hall.
Gaines, G. L. (1966). Insoluble Monolayers at Lipid-Gas Interfaces, Interscience,
New York.
Gege C, Schneider MF,et al. (2003) Artificial Rafts of Glycolipids with Partially
Fluorinated Membrane Anchors � Impact on Cell Adhesion. Submitted
Geyer, A., C. Gege, et al. (1999). �Carbohydrate-Carbohydrate Recognition
between LewisX Glycoconjugates.� Angew Chem Int Ed Engl 38(10): 1466 -
1468.
Geyer, A., C. Gege, et al. (2000). �Calcium-Dependent Carbohydrate-
Carbohydrate Recognition between Lewis(X) Blood Group Antigens. Angew Chem Int Ed Engl 39(18): 3245-3249.
Giorgione, J. R., Z. Huang, et al. (1998). �Increased activation of protein kinase C
with cubic phase lipid compared with liposomes.� Biochemistry 37(8): 2384-
92.
Goldenfeld, N. (1992). Lectures on Phase Transitions and the Renormalization
Group. Massachusetts, Addiso-Wesley Pubishing:.
Hakomori, S. I. (1991). �Carbohydrate-Carbohydrate Interaction as an Initial Step
in Cell Recognition.� Pure & Appl. Chem. 63(4): 473 - 482.
Harper, P. E., D. A. Mannock, et al. (2001). � X-ray diffraction structures of some
phosphatidylethanolamine lamellar and inverted hexagonal phases.� Biophys J 81: 2693 - 2706.
Harris, J. M. (1992). Poly(ethyleneglycol) Chemistry. New York.
Hato, M. and H. Minamikawa (1996). Langmuir 12: 1658.
Hinz, H. J., H. Kuttenreich, et al. (1991). �Stereochemistry and size of sugar head
groups determine structure and phase behavior of glycolipid membranes: densitometric, calorimetric, and X-ray studies.� Biochemistry 30(21): 5125-38.
Literature
99
Jacobson, K. and C. Dietrich (1999). �Looking at lipid rafts?� Trends Cell Biol
9(3): 87-91.
Keller, D. J., H. M. McConnell, et al. (1986). �Theory of Superstructures in Lipid
Monolayer Phase Transition.� J. Phys. Chem. 90: 2311 - 2315.
Köberl, M., H. J. Hinz, et al. (1998). �Temperature scanning simultaneous small-
and wide-angle X-ray scattering studies of glycolipid vesicles: areas, expansion coefficients and hydration.� Chem Phys Lipids 91: 13 - 37.
Kraegel, J., G. Kretzschmar, et al. (1996). �Surface rheology of monolayers.�
Elsevier. Thin Solid Films 284(285): 361-4.
Landau, L. D. and E. M. Lifschitz (1987). Lehrbuch der Theoretischen Physik.
Berlin.
Landersjö, C., C. Höög, et al. (2000). J Phys Chem B 104: 5618.
Larrson, K. (1988). Physical properties - structural and physical characteristics.
The Lipid Handbook. J. L. H. F. D. Gunstone, and F. B. Padley. London, Chapman and Hall: 321 - 337.
Lipowsky, R. (1991). �The conformation of membranes.� Nature 349: 475 - 481.
Maffettone, P. L., M. Grosso, et al. (1996). �Extensional Flow of a Two-
Dimensional Polymer Liquid Crystal.� Macromolecules 29: 8473-8478.
Mathe, G., A. Albersdorfer, et al. (1999). �Disjoining Pressure and Swelling
Dynamics of Thin Adsorbed Polymer Films under Controlled Hydration
Conditions.� Langmuir 15: 8726 - 8735.
Mathe, G., C. Gege, et al. (2000). �Equilibrium Swelling Behavior of Solid
Supported Poly(ethylene glycol) Lipid Monolayers. Effects of Short Chain
Lengths.� Langmuir 16: 3835 - 3845.
McConnell, H. M. (1991). �Structures and Transitions in Lipid Monolayer at
the Air-Water Interface.� Annu Rev Phys Chem 42: 171 - 95.
Miller, A. and H. Mohwald (1987). �Diffusion limited growth of crystalline domains in phospholipid monolayers.� Journal of Chemical Physics 86(7): 4258-65.
Literature
100
Möhwald, H. (1995). Structure and Dynamics of Membranes., Elsevier:
Amsterdam.
Morath, S., A. Stadelmaier, et al. (2002). �Synthetic lipoteichoic acid from
Staphylococcus aureus is a potent stimulus of cytokine release.� J Exp Med 195(12): 1635-40.
Mouritsen, O. G. (1983). �Studies on the lack of cooperativity in the melting of lipid
bilayers.� Biochim Biophys Acta 731(2): 217-21.
Naumann, C. A., C. F. Brooks, et al. (1999). �Viscoelastic Properties of
Lipopolymers at the Air-Water Interface: A Combined Interfacial Stress
Rheometer and Film Balance Study.� 15: 7752-7761.
Oliveira Jr, O. N., D. M. Taylor, et al. (1992). �Modelling the surface potential-area
dependence of a stearic acid monolayer.� Thin Solid Films 210/211: 76-78.
Pabst, G., M. Rappolt, et al. (2000). �Structural information from multilamellar
liposomes at full hydration: Full q-range fitting with high quality x-ray data.� Phys Rev E 52: 4000 - 4009.
Pallas, N. R. and B. A. Pethica (1985). Langmuir 1: 509.
Perez, S., N. Mouhous-Riou, et al. (1996). �Crystal and molecular structure of a
histo-blood group antigen involved in cell adhesion: the Lewis x
trisaccharide.� Glycobiology 6(5): 537-42.
Pozzoli, G., G. Vita, et al. (1997). Melt-Proccessable Perfluoropolymers. In Modern
Fluoropolymers., Wiley.
Riess, J. G. (2002). �Fluorous micro- and nanophases with a biomedical
perspective [Review].� Tetrahedron 58(20): 4113-4131.
Riess, J. G. and J. Greiner (2000). �Carbohydrate- and related polyol-derived
fluorosurfactants: an update.� Carbohydr Res 327(1-2): 147-68.
Rundlöf, T., C. Landersjö, et al. (1998). Magn. Reson. Chem. 36: 773.
Sackmann, E. (1996). Structure and Dynamics of Membranes. From Vesicles to
Cells. E. Sackmann and R. Lipowsky. Elsevier: Amsterdam.
Literature
101
Sano, Y., T. Sasaki, et al. (1991). Photon Factory Activity Rep. 9: 196.
Schneider M.F., Andelman D. and Tanaka M. (2003) Electrostatically Driven Stripe
Phase Formation of Fluorinated Lipid Monolayers at the Air/Water Interface.
In Preparation.
Schneider MF, Marsh D et al. (1999) Network formation of lipid membranes:
triggering structural transitions by chain melting. Proc Natl Acad Sci U S A
96(25):14312-7.
Schneider, M.F., Lim, K., et al. (2002). "Rheology of Glycocalix Model at Air/Water
Interface." PCCP 4(10):1945-52
Schneider, M. F., G. Mathe, et al. (2001). �Thermodynamic Properties and
Swelling Behavior of Glycolipid Monolayers at Interfaces.� J. Phys. Chem. B 105: 5178.
Schneider, M. F., R. Zantl, et al. (2002). �Hydrophilic/Hydrophobic Balance
Determines Morphology of Glycolipids with Oligolactose Head Groups.�
Biophysical Journal.
Scott, H. L., Jr. (1975). �Some models for lipid bilayer and biomembrane phase
transitions.� Journal of Chemical Physics 62(4): 1347-53.
Seddon, J. M., G. Cevc, et al. (1984). �X-ray diffraction study of the polymorphism
of hydrated diacyl- and dialkylphosphatidylethanolamines.� Biochemistry 23:
2634 - 2644.
Shin, S., N. Collazo, et al. (1993). �Comment on molecular dynamics simulations
of monolayers of fluorinated amphiphiles.� Journal of Chemical Physics 98(4): 3469-74.
Simons, K. and E. Ikonen (1997). �Functional rafts in cell membranes.� Nature
387(6633): 569-72.
Smondyrev, A. M. and M. L. Berkowitz (1999). �Molecular dynamics simulation of
fluorination effects on a phospholipid bilayer.� Journal of Chemical Physics
111(21): 9864-70.
Literature
102
Tamada, K., H. Minamikawa, et al. (1996). Langmuir 12: 1666.
Vogel, J., G. Bendas, et al. (1998). �The role of glycolipids in mediating cell
adhesion: a flow chamber study.� Biochim Biophys Acta 1372(2): 205-15.
Vogel, V. and D. Mobius (1988). �Hydrated polar groups in lipid monolayers:
effective local dipole moments and dielectric properties.� Thin Solid Films
159: 73-81.
Warren, B. E. (1969). X-ray diffraction. Reading, MA, Addison-Wesley.
Weis, R. M. and H. M. McConnell (1985). �Cholesterol stabilizes the crystal-liquid
interface in phospholipid monolayers.� Journal of Physical Chemistry 89(21):
4453-9.
Wiener, M. C., R. M. Suter, et al. (1989). �Structure of the fully hydrated gel phase
of dipalmitoylphosphatidylcholine.� Biophys J 55: 315 - 325.
Literature
103
Publications
1) Winter DC, Schneider MF, O�Sullivan GC, Harvey and Geibel JP Rapid effects
of aldosteron on sodium-hydrogen exchange in isolated colonic crypts. J.
Membrane Biol. (1999) 170, 17-26.
2) Schneider MF, Marsh D, Jahn W, Kloesgen B, Heimburg T. Network formation
of lipid membranes: triggering structural transitions by chain melting. Proc Natl
Acad Sci U S A (1999) Dec 7;96(25):14312-7.
-------------------------------------------------------------------------------------------------------------
3) Schneider MF, Mathe G, Tanaka M, Gege C, Schmidt RR (2001)
Thermodynamic properties and swelling behavior of glycolipid monolayers at
interfaces. J. Phys. Chem. B 105:5178-5185.
4) Schneider MF, Lim K, Fuller GG and Tanaka M (2001) Rheology of Glycocalix
Model at Air/Water Interface.PCCP 4(10):1945-52
5) Schneider MF, Zantl R, Gege C, Schmidt RR and Tanaka M (2002)
Hydrophilic/Hydrophobic Balance Determines Morphology of Glycolipids with
Oligolactose Head Groups. Biophys. J., In Press
6) Gege C, Schneider MF, Schumacher G.,Limozin L, Rothe U,Bendas G, Tanaka
M, and Schmidt, RR (2003) Artificial Rafts of Glycolipids with Partially Fluorinated
Membrane Anchors � Impact on Cell Adhesion. Submitted
7) Schneider MF, Andelman D and Tanaka M (2003) Electrostatically Driven Stripe
Phase Formation of Fluorinated Lipid Monolayers at the Air/Water Interface. In
Preparation.
8) Vautrin V, Zemb T, Schneider MF, Tanaka M (2003) Balance of pH and Ionic
Strength Influences on chain melting transition in catanionic vesicles. Submitted
9)Tanaka M, Schneider MF, Brezesinski G (2003) In-planeMorphology of
Synthetic Oligolactose Lipid Monolayers � Impact of Saccharide Chain Length.
ChemPhysChem, In Press.
104
CURICULUM VITAE
MATTHIAS F. SCHNEIDER
PERSONAL INFORMATION
! Date of Birth : 02. Feb 1971 ! Place of Birth : Schweinfurt, Bayern, Germany
UNIVERSITY EDUCATION
2000 - Lehrstuhl für Biophysik TU München Promotion (PhD- Thesis) ! Physik der Gylkokalix (Physics of the Glycocalix)
1995 - 1999 Physik Fakultät Universität Göttingen Student ! Physik Hauptdiplom
! Diplomarbeit at the Max Planck Institut for biophysikalische Chemie, Title : “Thermodynamik von Membranen und Membrannetzwerken“.
! Diplomprüfung : Oktober 1999 (Score : 1,6)
1994 - 1995 Physik Fakultät Universität - GH Siegen Student ! Physik Grundstudium
! Vordiplom : August 1995 (Score: 2,0)
1993 - 1994 Physikalische Technik FH - Rüsselsheim Student ! Grundstudium
ADDITIONAL RESEARCH EXPERIENCE
1997 Yale Medical School New Haven, CT, USA Guest Scientist ! Department of Physiology, Prof. Dr. John Geibl
! Subject : “Rapid Effects of Aldosteron on Sodium-Hydrogen Exchange using Laser Confocal Microscopy”
! From : March – November (9 Month)
2001 Stanford University Stanford, CA, USA Visiting Scientist ! Department of Chemical Engineering (Prof. Dr. Gerry. G. Fuller)
! Subject : “Rheology of Glycolipid Monolayer“
! May 2001 (5 Weeks)
105
CIVIL SERVICE (ERSATZDIENST)
1991 – 1992 Leopoldina Krankenhaus Schweinfurt, Bayern, Germany.
PRIMARY EDUCATION
1989 – 1992 Abitur, FOS Schweinfurt, Bayern (Technischer Zweig)
1982 – 1989 Gymnasium Bad Königshofen (Rhön Grabfeld)
1978 – 1982 Grundschule Stadtlauringen (Ldkr Schweinfurt)